Semantic learning designs: recording assumptions and guidelines
-
Upload
miguel-angel-sicilia -
Category
Documents
-
view
216 -
download
3
Transcript of Semantic learning designs: recording assumptions and guidelines
copy 2006 The Author Journal compilation copy 2006 British Educational Communications and Technology Agency Published by Blackwell Publishing 9600 Garsington Road Oxford OX4 2DQ UK and 350 Main Street Malden MA 02148 USA
British Journal of Educational Technology Vol 37 No 3 2006
331ndash350doi101111j1467-8535200600609x
Blackwell Publishing LtdOxford UKBJETBritish Journal of Educational Technology0007-1013British Educational Communications and Technology Agency 20062006373331350Articles
Semantic learning designsBritish
Journal of Educational Technology
Semantic learning designs recording assumptions and guidelines
Miguel-Angel Sicilia
Miguel-Angel Sicilia is at the University of Alcalaacute Spain Dr Miguel A Sicilia leads the InformationEngineering Research Unit at the Computer Science Department University of Alcalaacute His researchinterests are primarily in the areas of adaptive hypermedia learning technology and human-computerinteraction He is coauthor of more than twenty refereed journal publications in these areas and he alsoleads the AIS SIG on reusable learning objects (httpwwwsigrloorg) Address for correspondenceComputer Science Department University of Alcalaacute Ctra Barcelona km 336 Madrid 28871 SpainEmail msiciliauahes
Abstract
Recent developments in the standardisation of learning technology haveresulted in models of learning activities and resources including descriptivemetadata and definitions of conditional flows for multirole activitiesNonetheless such
learning designs
are actually representations of the
results
ofthe design process and do not provide information about the
rationale
of thedesign ie about the theoretical standpoints assumptions or guidelines appliedto come up with the concrete arrangement of activities These latter elementsare critical not only for informative reasons but as a medium towards the endof connecting theories and hypotheses to actual practice and analysing theresulting empirical data as a form of inquiry on the validity of theoreticalassumptions This paper delineates the main aspects of a schema for therecording of such design rationales using an ontological approach The methodfor the engineering of the schema was based on connecting the definitionsprovided with an existing large ontological base thus reusing a large amountof common sense knowledge Two paradigmatic example positions of the rangeof aspects that could be covered by the representation language are describedas an illustration The resulting ontological definitions can be used as afoundation for the refinement of theoretical positions and for their comparativeassessment
Introduction
The recent evolution and growth of online learning activities of a diverse kind hasresulted in new practices and conceptual paradigms for the crafting of learningresources and learning programs This includes specific knowledge that applies to the
design
of e-learning materials and activities For example Brown and Voltz (2005) haverecently summarised some of them ranging from technological to social or cognitive
332
British Journal of Educational Technology Vol 37 No 3 2006
copy 2006 The Author Journal compilation copy 2006 British Educational Communications and Technology Agency
and Conole Dyke Oliver and Seale (2004) have provided a model for designing e-learning activities from different theoretical positions In addition learning technologystandards that model learning contents and activities (Friesen 2005) provide enhancedopportunities for the reuse of diverse kinds of resources stemming from the concept oflearning object (Polsani 2003 Wiley 2001) These resources range from self-standingeducational materials to complex learning activity designs encompassing the interac-tion of several roles and being driven by rules explicitly provided to guide the flow ofactivities
Concretely the increasing adoption of the
IMS Learning Design
(LD) specification(httpwwwimsprojectorg) has provided a standardised common ground to therepresentation of learning designs in digital form An
activity
in IMS LD is consideredas a piece of interaction among a number of specified rolesmdashplayed by persons ininstantiations of each concrete learning designmdashthat produce a tangible outcome byusing a concrete environment The environment is in turn made up of
learning objects
(McGreal 2004) and
services
(ie facilities available at runtime as for example a chatroom) Activities can be further decomposed in subactivities and they are aggregatedinto
methods
that specify the conditions for application of the learning design alongwith the planned objectives that will eventually match the outcomes of the activitiesMethods can be structured around concurrent
plays
and these in turn can be structuredin sequential
acts
the latter allowing the specification of conditional execution Thisschematic description of learning design gives an idea of the flexibility the specificationprovides in describing activity-based learning programs as acknowledged elsewhere(Allert 2004) IMS LD combined with learning object-oriented specifications suchas IEEE Learning Object Metadata (LOM) (IEEE 2002) and ADL-SCORM (httpwwwadlnetorg) enable a degree of representation of activities and materials withunprecedented applications
As can be appreciated from the earlier discussion there exists a growing consensus ona concrete terminology and model for creating metadata describing learning activitiesand their associated constituents and at least general principles for their design are yetavailable Nonetheless while this is enough for describing the activities by themselvesa richer framework is required to capture also the intellectual process that was involvedin their crafting ie the hypotheses assumptions and decisions that led to the concretearrangement of activities and conditions Doing so will expand the current scope ofmetadata-described learning resources from the description of the activities themselvesto the description of the designerrsquos intentions It may be argued that such kind ofknowledge is difficult to capture because of the heterogeneity of theoretical standpointsand also because of the use of lsquotacitrsquo knowledge or little reflexive designs that could befound in many designs But the benefits available for researchers and practitioners ingeneral are still important enough to deserve the effort and in fact toolkits as the onedescribed by Conole
et al
(2004) consider design theories explicitly and can be used asa point of departure The benefits of recording the rationale for learning designs iscomplementary to the
postactivity
analysis of the actual history and outcomes of activ-ities in actual settings which has been pointed out as a relevant source of empiricalinformation (Koper 2004) that will eventually be available in the coming years
Semantic learning designs
333
copy 2006 The Author Journal compilation copy 2006 British Educational Communications and Technology Agency
More concretely the modelling of the rationale of learning designs would provide anumber of benefits for researchers and educators in general Among these benefitsSicilia and Lytras (2005) provided the following list (1) the
linking of theoretical assump-tions to practical learning designs
for informative purposes (2) the use of such
links asresources for education of learning designers
(3) the search for
patterns in design situations
that include some given theoretical assumptions and (4) the lsquoinferencersquo of
detailed com-parisons between the effectiveness and adequacy of learning designs
(for similar settings)These aspects are reflections of the practical benefits of the main objective of the record-ing of design rationales which is no other than advancing the scientific inquiry onlearning by enabling the recording of explicit links from design rules to concretedesigns and the latter to the outcomes of the lsquoinstantiationrsquo of these designs in possiblydiverse institutions or groups This would come up with an unprecedented empiricalbase for meta-analysis and knowledge discovery For example hypotheses on the ade-quacy of group-based or individual learning for some given conditions could be testedwith a large empirical base containing the history of interaction and results of multiplesimilar learning programs
This paper approaches the problem of the metadata representation of assumptions andhypothesis related to learning designs in a general sense Because the entities to berepresented are relevant rules or rationales involved in the intellectual process of design-ing learning experiences the schema described is deliberately open-ended This is arequirement imposed by the fact that theories and guidelines related to learning arebased on models that are abstract in nature This in turn is because they attempt tocharacterise learning as a general human cognitive process but there exist at least somerecurring elements that are common to many practical learning designs We have notattempted to build an ontology of existing theories of learning but addressing somelsquoupperrsquo elements that cross-cut any of them eg notions of
change
are one of these upperelements (Sicilia amp Lytras 2005) In any case the fact that diverging accounts oflearning exist is far from controversial and even modern theories somewhat diverge intheir basic ontological commitments (Packer amp Goicoechea 2000)
The representation described departs from a model based on IMS LD but expressed in aricher ontological language Ontology engineering (Gruber 1995) provides a methodto develop semantic conceptualisations and
Semantic Web
technology (Berners-LeeHendler amp Lassila 2001) aims at providing tools and techniques to develop softwarethat exploits them in the context of a learning system Formal ontology provides theknowledge representation infrastructure for ontologies of learning to a level of consid-erable complexity and richness as supported by description logics (Baader CalvaneseMcGuinness Nardi amp Patel-Schneider 2003) It should be noted that the lsquoontology oflearningrsquo explicitly or tacitly assumed actually constraints and drives the behaviourof the system as a whole and the criteria used for the evaluation of the outcomes ofactivities also become a reflection of previous ontological commitments For examplean ontology not considering social issues will not allow (or will not care about) thetracking of the evolution that takes place in the social network of the learners involvedas the activities are carried out
334
British Journal of Educational Technology Vol 37 No 3 2006
copy 2006 The Author Journal compilation copy 2006 British Educational Communications and Technology Agency
The resulting learning design descriptions are called here lsquo
semantic learning design
rsquo torepresent the fact that they are described in ontological terms and that the meaning ofthe design as conceived by the designer is explicitly represented This carries one stepfurther the current metadata descriptions provided by specifications such as LOMSCORM or LD
As a technique for validating the semantic precision of the schema described hereexplicit links have been provided to concepts and relations that are already described ina large upper ontology concretely the
OpenCyc
09 knowledge base This is an alterna-tive to analysis techniques such as the BungendashWandndashWeber (Wand amp Weber 1995)that fosters the reuse of existing open knowledge engineering The subsequent mappingto modern Web-enabled ontology languages such as OWL Web Ontology Language(httpwwww3orgTRowl-ref) is a straightforward step In addition some represen-tative examples of different theoretical standpoints are described as illustrations
The rest of this paper is structured as follows Section 2 provides an ontological accountof an activity-based model of learning activities as supported by IMS LD which consti-tutes the
artefact
that was created by an intellectual process Then Section 3 describesa proposal for an ontological schema targeted at recording design rationales in a broadsense The scheme described is then used in the fourth section to provide as a case studythe comparison of two theoretical frameworks that lead to two different descriptions fora concrete learning design Finally conclusions and future research directions are pro-vided in the fifth section
Recording educational metadata and semantic annotation of learning activities
The first step in the modelling of learning design rationales is that of representing theresults or
artefacts
of the design process In this section the main elements of the IMSLD schema are expressed in ontological terms considering both the design in itself (iethe equivalent to a learning design description) and the entities that result from theapplication of designs in actual occurrences of learning following them (ie the record-ing of actual executions of the planned design in concrete instructional or trainingsettings)
A significant amount of reuse in terminological structures can be achieved by buildingconceptualisations on top of existing large terminological bases like
OpenCyc
(httpwwwopencycorg)
OpenCyc
is the open source version of the
Cyc
knowledge base(Lenat 1995) which contains over 100 000 atomic terms and is provided with anassociated efficient inference engine
Cyc
uses as its underlying definition language avariant of predicate calculus called
CycL
and it attempts to provide a comprehensiveupper ontology of lsquocommon sensersquo knowledge In the rest of the paper the ontologicaldefinitions provided are explicitly linked to
OpenCyc
definitions (in which ontologicaldefinitions of terms and relations are provided in courier font and those that are definedin
OpenCyc
are prefixed by lsquoocrsquo) as a means of reusing existing ontological engineeringefforts and also as a way of validating the concepts being represented
Semantic learning designs
335
copy 2006 The Author Journal compilation copy 2006 British Educational Communications and Technology Agency
Describing the main structural elements of IMS LD
The conceptual model of IMS LD (IMS 2003) provides a concise lsquometalanguagersquo thatcould capture a diversity of pedagogical approaches In doing so it essentially adheresto an activity-centred paradigm (Allert 2004) It should be noted that this model isneutral to specific learning theories Even though it integrates the hierarchical struc-ture of activities and mediation principles of
Activity Theory
(Kaptelinin Kuutti amp Ban-non 1995) it is actually a metalanguage that does not make assumptions about otherelements of such theory as the internalisationexternalisation differentiation
The first step in developing an ontological schema for IMS LD must be that of clarifyingthe main entities to be represented
OpenCyc
provides a rich framework to describetemporal activities concretely the
oc_PurposefulAction
concept (a kind of
oc_Event
) defined as lsquoan action consciously volitionally and purposefully done by atleast one actorrsquo captures the essence of learning activities which are the ultimate targetof the specification Nonetheless IMS LD metadata does not attempt to representing theactivities as actually enacted by persons but they describe
blueprints
that will eventuallybe bound to one or several instantiations of activities and they are intended to besearched as such (Buzza Bean Harrigan amp Carey 2005) In contrast
oc_Event
s andtheir associated entities are related to instantiations and not to their specificationsFurther the concept of
oc_ActorSlot
represents the specific roles played by personsin events so that this could be used to represent the LD
staff
and
learner
role categorybut they do not represent roles understood as templates for their eventual binding toconcrete people
An alternative to representing LD specifications as blueprints is that of creating
Event
types for each activity defined in a learning design But this approach entails at leasttwo problems On the one hand this is not ontologically correct because it does notrepresent the reality that the production of IMS LD is an act of creating specificationartefacts instead of a process of classification of possible real world entities In factmany learning designs will probably be never executed and some of them may bepurposefully created as mere design examples On the other hand the direct creationof learning designs as ontology terms entails that the activities actually performed mustadhere strictly to the original learning design used which severely restricts the flexibil-ity in using the design with ad hoc changes or modifications (or even modified
versions
)This second practical flaw suggests that LD specifications should be represented asseparated entities
This has led us to the major decision of modelling learning designs as specificationsfollowing the definition of
oc_Specification
as lsquoan abstract work that constitutesa description of the properties of a
Situation
or a
SomethingExisting
and sometimes evenentire collections of such thingsrsquo Table 1 summarises the representation of
LearningDesigns
in terms of specifications (specialisations of
oc_Specification
) their mainrelated properties and the implications of each of the specifications when producingactual activities Notifications and completion conditions are omitted for brevityThe concepts of lsquo
learning design
rsquo and lsquo
method
rsquo can be considered as equivalent from an
336
British Journal of Educational Technology Vol 37 No 3 2006
copy 2006 The Author Journal compilation copy 2006 British Educational Communications and Technology Agency
Table 1 Summary of the main specification definitions in the ontological schema
Concept Main descriptions Consequences in actual activities
Method-Spec Objectives and prerequisites(described in the followingsection) Structured in
play
specifications or directly
act
specifications
Instantiation is an independentmeaningful learning activity Playsare cotemporal
oc_subEvent
ssubsumed in the method Note thatthe semantic definition of subeventsyet entails the subsuming of subeventsin the containing event
Play-Spec Collection of sequential acts thatcomprise the play
Because plays are potentiallyconcurrent there are no furthertemporal restrictions
Act-Spec Collection of concurrent role parts that detail the activitiesincluded in the design Role partsand environments are the mainelements related to acts
The sequence of instantiations of actscould be sequenced through the
oc_startsAfterEndingOf
RolendashPart-Spec The linking of role specificationsto concrete activities or activitystructures
This entity is not required in theontological schema because theparticipation of roles in the activitiescan be stated in terms of specificationsof
oc_ActorRole
s or subeventsthat are not of a temporal nature
Activity-Spec The specification of learning andsupport structures in terms of
Role-Spec
s
ibtUsed LearningObject
s and used
Service
s Concrete objectivescan also be attached
Each instantiation of an
Activity-Spec
will result ininstances of specific types of
oc_Activity
Following LD specs
support
activities instantiate for everyrole mapping while
learning
onesinstantiate for all the role mappings
Environment These are descriptions of learning objects and services that do not require anexplicit mapping in the ontological schema
Service Service facilities have to belsquoinstantiated by a local run-timeservicesrsquo eg a chat or discussionforum Their computationalnature can be described interms of concrete
oc_ComputerProgramCW
Subcategories of such conceptualrepresentations of softwareentities can be used for thetypology of services
The
oc_ComputerProcessRunning
term is used to describe executionsof
oc_ComputerProgramCW
which cover every instance of aservice as understoodin IMS LD Thelinking between conceptual andactual running instances is coveredby the required predicate
oc-programOfProcess
Semantic learning designs
337
copy 2006 The Author Journal compilation copy 2006 British Educational Communications and Technology Agency
Learning object Those descriptions provided bylearning object metadataLearning objects can be definedas
oc_IBT
s as described inSicilia Lytras Rodriacuteguez andGarciacutea (in press) and approachesto describing metadata throughontologies yet existmdashsee an overview in Sicilia and Garciacutea(2005)
The learning objects selected arerepresented as instances in theontology which are declared to beused in activities through the
ibtUsed
predicate
Role-Spec Describe stereotypical actors oflearning and support activitiesThese act as lsquocontractrsquo for theactual participants in the senseprovided by Saacutenchez-Alonso andSicilia (2004)
Role specs are mapped to actualusers when the activities take placeThis is mapped to predicates of theclass
oc_ActorRole
Concept Main descriptions Consequences in actual activities
LD
=
Learning Design
Table 1 Continued
ontological perspective as the former allow onemdashand only onemdashinstance of the latterwith no particular logical differences
As described in Table 1 the
OpenCyc
built-in support for modelling temporal events canbe used to express the coordination relationships between the constituents of the unitsof learning The actual instances of learning events will be modelled as instances of
oc_Learning
(lsquothe collection of all events brief or extended in which an agent isacquiring information or know-howrsquo) and more specifically as instances of a speciali-sation
ComputerMediatedLearning
simply requiring the use of information net-works for at least part of the learning activities Because events in
OpenCyc
are alsosituations they have components in addition to the temporal extent including peopleinvolved information things used and other arrangements
Figure 1b depicts the main elements that describe actual current or past enactmentsof activities The schema is based on a clear separation of the specifications and theactual activities that by itself constitutes a metadata layering
OpenCyc
predicatesas
oc_situationConformsToSpecification
allows the linking of situations(every action is a situation also) to the specifications they conformmdashbe it purposely ornot The
LD-Activity term has been introduced as the common category of learningand support activities Methods plays and acts become special subcategories of LD-Activity and the constituent activities of Acts are instances of LD Activity also Rolesare attached to any kind of LD Activity even though it is required for the low-levelstructure The learning design concept of lsquoactivity structurersquo as an aggregation ofactivities does not require an explicit modelling because of the generic mereology ofactivities engineered in OpenCyc
338 British Journal of Educational Technology Vol 37 No 3 2006
copy 2006 The Author Journal compilation copy 2006 British Educational Communications and Technology Agency
Figure 1 (a) Summary of the competency components in the ontological schema (b) Summary of the main activityndashrole definitions in the ontological schema
oc_WorkingEvent
oc_IntelligentAgent
The competency elements aremeasurable characteristics andthey will be typed by the relatedSpecification classes eg Skill
SpecputlnPracticeln
(a)
Competency Competency-Spec
Competencyelement
Skill oc_ Competence
Attitude
Knowledge
has
thingSpecified
oc_Learning
ComputerMediatedLearning
oc_Role
(b)
Staff
Learner
oc_Specification
Method-Spec
Play-Spec
Act-Spec
Support
LD-Activity
contains
contains
Method
Play
Act
startAfterEndingOf
subAction
subAction
Activity-Spec
ltltActorSlotsgtgt
situationConformsToSpec
Activity-Spec
Activity
Semantic learning designs 339
copy 2006 The Author Journal compilation copy 2006 British Educational Communications and Technology Agency
The concept of oc_Role accounts not only for participant roles but also for otherkinds of lsquocomponentsrsquo This is why two specialisations of the concept ofoc_ActorSlot labelled Staff and Learner are required to differentiate partici-pants from other elements As is obvious from the earlier discussion the built-in rep-resentation of activities in OpenCyc is powerful enough for the definition of learningdesign activities
The final part of the ontological specification described so far is that of the mappingfrom LD specifications to the actual past or present activity realisations Theoc_situationConformsToSpecification predicate can be used to trace whichactivity realizations come from a given learning design activity definition and the sameoccurs with the oc_thingSpecified for other learning design elements such asroles or environments Existing learning design execution engines as CopperCore(httpcoppercoreorg) actually map LD specifications to online collaboration so thatthe translation of the ontological representations of learning designs to ontology defi-nitions of activities is straightforward following similar representation
Describing objectives outcomes and preconditionsAccording to IMS learning design learning objectives and prerequisites lsquocan bedescribed using the IMS Reusable Definition of Competency or Educational Objective(RDCEO) format but can also refer to simple resources (eg a text) with a descriptionof the learning objectiversquo As the latter option is not useful from the viewpoint of com-putational semantics we will adhere to the former As commented earlier the compe-tency view is not exclusive to other kinds of objectives related to other ontologicalstandpoints (Sicilia amp Lytras 2005) and other conceptions of change or improvementmay include social ties cognitive structures or the combination of the knowledge of agroup In fact competencies adhere mostly to behaviouristic frameworks that focus onobservable external behaviour (Tennant 1991) rather than on cognitive representa-tions Nonetheless the further discussion is equally applicable to other kinds of objec-tives so that we will adhere to the competency view
oc_Competence in OpenCyc is defined as a quantity that lsquofor describing the level ofskill with which an agent performs some taskrsquo This does not remove the lack of preci-sion that is yet present in RDCEO definitions as described by Sicilia (2005) As analternative Figure 1a provides a simplified account of the model described by Sicilia asrelated to OpenCyc terms in which competence is integrated as one of the elements of thebroader notion of competency according to the model of Rummel (Rothwell amp Kazanas1992) which considers work situations attributes of the individual that are applied insuch situations and measurement with respect to some standard of performance Themost relevant distinction is again the separation of actual competency representationsand competency lsquodefinitionsrsquo (ie specifications) intended to provide a type structure tocompetencies to be used in objective or precondition specification In this case thegeneric oc-thingSpecified predicate can be used for the linking of both levels ofrepresentation
340 British Journal of Educational Technology Vol 37 No 3 2006
copy 2006 The Author Journal compilation copy 2006 British Educational Communications and Technology Agency
Objectives and preconditions can then be expressed as formulas on the acquisition ofcompetencies or their components (competency elements) Descriptions specific toOpenCyc may use oc_CycLSentence-Assertible but simpler enumeration ofcompetency components could be used instead
Describing learning design rationales theories hypotheses guidelines and rulesThe concept of lsquodesignrsquo for our purposes is properly captured in the 5a definitionprovided in the Merriam Webster Online Dictionary (wwwm-wcom) lsquoan underlyingscheme that governs functioning developing or unfoldingrsquo In consequence learningdesigns refer to artefacts and not to the process of devising the educational solutionThis is equivalent to the concept oc_Design subsumed by the notion ofoc_Specification which is described as lsquoan abstract work that constitutes adescription of the properties of a Situation or a SomethingExisting and sometimeseven entire collections of such thingsrsquo In consequence the term LearningDesignwould simply delineate a subset of specifications that are oriented to describe howlearning experiences with concrete objectives should be Trivially oc_Designerdescribes the authors of the learning designs and oc_Designing represent theoc_HumanActivity that leads to the production of learning designs Designing as aspecial kind of activity is carried out (oc_doneBy) by a designer which provides a wayto attach metainformation about the rationale of the design
From an ontological perspective the departing problem in modelling learning designrationales is that their assumptions and hypotheses of different theoretical standpointsneed not be compatible or need not consider the same aspects of reality (Conole et al2004 Packer amp Goicoechea 2000) Furthermore different pedagogical approaches areknown to result in different mappings to activities (Conole et al 2004) This wouldeither require the provision of separate ontologies or the use of a representationalmechanism that allows such kind of potential inconsistency or divergence The conceptof microtheory in OpenCyc provides such representational mechanism intended to orga-nise assertions that depend on lsquoshared set of assumptions on which the truth of theassertions dependsrsquo Definitions inside the same microtheory need to be consistent butthis is not required across microtheories The descriptions provided in the earlier sectioncan be used as a base microtheory called LearningActivitiesMt for others that are spe-cific to concrete theoretical frameworks In this section we will describe elements thatare common to the description of theoretical frameworks the distinction among micro-theories will not be discussed
The intellectual processes of learning design in themselves will be represented asoc_Designing activities carried out by oc_Designers An additional consider-ation is that the information about the rationale of the design would be better attachedto representations of each of the design processes (ie to oc_Designing instances)than to the final design This enables the registering of different assumptions for differ-ent contributions to the same artefact eg different revisions from the same of differentdesigners
Semantic learning designs 341
copy 2006 The Author Journal compilation copy 2006 British Educational Communications and Technology Agency
Several designers may participate in the same designing activity and formal frameworksfor collaborative learning resource creation (Dodero Aedo amp Diacuteaz 2002) can be inte-grated by clearly specifying which class of things can be the outcome of a designingprocess For the sake of flexibility the concept of LearningDesignPart is introduced as anall-encompassing category that includes each of the elements described in the previoussection This includes not only activities and roles but also objectives conditions andthe like Furthermore the granularity of the descriptions allows for different levels fromconsidering a unique designing activity for the whole package to recording each ele-ment of the learning design comprehensively Because oc_Actions are subsumed bythe concept of oc_Event considered as a lsquodynamic situation in which the state of theworld changesrsquo each edition of a metadata record or any other form of description maybe subject to be described The concept is flexible enough to be able to record fine-grained detailed decisions This of course may include different versions of the sameartefacts but this is an orthogonal aspect of configuration that is not relevant to ourpresent discussion
Figure 2 depicts the main elements of the loci to which rationales assumptions andguidelines will be attached
A LearningDesigner can be defined as the class of humans that performactivities of learning resource design The consideration of learning design as aunique human activity could be considered as restrictive of intelligent softwarefor the task Simply changing the oc_Designer class to a subconcept ofoc_IntelligentAgent could remove this but this entails a change in Open-Cyc knowledge base The concept oc_Designing is a subclass of the conceptoc_ThinkingAboutTheStateOfTheWorld which represents lsquothinking pro-cessesrsquo which include mental events such as planning evaluating or inventingThus the concept of design described so far should be understood in terms of the
Figure 2 Main elements of the ontological schema for the loci of design rationales
oc_performedBy modifies
342 British Journal of Educational Technology Vol 37 No 3 2006
copy 2006 The Author Journal compilation copy 2006 British Educational Communications and Technology Agency
concept of lsquoexpanded rationalityrsquo as described by Hatchuel (2002) integrating cre-ativity and unexpected expansions of the original requirements This precludes onto-logical definitions in which the problem space is completely bounded a priori Inconsequence a degree of openness is necessary to integrate different kinds of detailin description from fully described ones to others with shallower semantics eg someproviding only references to generic assumptions
Theoretical frameworksWe will make the distinction that guidelines (as further defined) are the propositions thatdirectly predicate about the validity or appropriateness of learning design parts whileother theoretical propositions about learning that are not directly saying anythingabout the design of digital learning resources are considered as assumptions followingthe general sense provided in the Merriam-Webster dictionary lsquoa fact or statement (asa proposition axiom postulate or notion) taken for grantedrsquo
The fact that some assumption hypotheses or proposed law about learning is sustainedby a broader empirical base or is better sustained by rational means is out of the scopeof this work because here we focus on design practice (even though it is of an obviousinterest for researchers) In principle even unorthodox positions may be sustainedbecause the mechanism of microtheories and their inclusion mechanism allows forgreat flexibility in modelling them
Table 2 provides some examples of theoretical assumptions and related guidelines forlsquoextremersquo archetypical standpoints on learning These are not intended to reflect actualcommon positions but they are provided as illustrative examples of radical positionsthat could be modelled with our ontological schema
The examples provided in Table 2 go a step further than toolkits in Conole et al (2004)in making the design rationale explicit Nonetheless in many cases lsquocommon sensersquo isapplied to design and a shallower approach to specification could be used instead egproviding simply links to the theoretical standpoints used It should be noted also thatthe resulting guidelines vary greatly in their level of concreteness and some of themmay be considered as mandatory from some standpoints while others are mere generalrepresentations This would lead to different computational strategies
Guidelines and rulesThe WordNet (httpwordnetprincetonedu) thesaurus defines guidelines as lsquoguidancerelative to setting standards or determining a course of actionrsquo or lsquoa rule or principlethat provides guidance to appropriate behaviorrsquo In our case guidelines are statementsthat serve as criteria for decision in designing learning Given that e-learning designis still far from being a mature discipline guidelinesmdashas instances of recordedknowledgemdashcan be characterised as oc_NormativeSpecifications becausethey describe how learning resource design activities and their outcomes lsquoshould bersquoin a propositional form Propositions lsquohave some truth value in some context or
Semantic learning designs 343
copy 2006 The Author Journal compilation copy 2006 British Educational Communications and Technology Agency
ldquoworldrdquo rsquo Furthermore propositions lsquoare assumed to be representable (at least in prin-ciple) by a sentence in some formal or natural languagersquo This raises the need for alanguage that expresses guidelines in a general sense There are two main options forsuch representation be them formal logical sentences or reified guidelines
Formal logical sentence representation entails the formalisation through sentences ofthe constraints of the designs For example if lsquoevery learning resource must be activersquoaccording to some theoretical standpoint (eg one that fosters reflective learning) thissentence may be represented as an axiom and put into the appropriate separate micro-theory as the following CycL fragment
($forAll X ($implies
($isa X $LearningObject)
($resourceType X $Active))
Then resources that are not active are simply excluded from the notion itself This is astrong position in the sense that precludes some views on the learning design process
Table 2 Some examples of theoretical assumptions and guidelines
Archetypicalstandpoint Assumptions Guidelines
Instructivist(Behaviouristic)
bull Learning occurs through association and reinforcement
bull The results of learning can be appreciated through observable outcomes
(GB1) Every instance of oc-Learning entails theexposure to some LearningObjectthat produces the association (GB2) Learning Activitysequences should provide assessmentactivities that confirm theassociations that should have beenestablished so far
Constructivist bull Learning proceeds by building mental structures
(GC1) Agentrsquos mental structuresshould be represented(GC2) These structures should bepre-assessed in order to becomparable with the resulting ones(GC3) Design Activities arerecommended(GC4) Contrast Activities arerecommended
Sociocultural bull Learning occurs through social participatio
(GSC1) Every instance ofoc-Learning entails theinteraction of severaloc-IntelligentAgents asoc-Actors
344 British Journal of Educational Technology Vol 37 No 3 2006
copy 2006 The Author Journal compilation copy 2006 British Educational Communications and Technology Agency
Another typical example may be that of a sociocultural guideline that preclude Methodsthat do not entail multirole activities that provide access to social interaction ServicesThis approach to guidelines provides the best framework for automation and wouldenable the creation of microtheories that represent clearly formulated views onlearning Nonetheless there are two major problems for the use of this option as theunique mechanism for encoding design rationales
1 There is no clear-cut distinction between models and theories of learning becausethey come from a history of schools of thought with interrelations (Palmer 2001)This would lead to the need for the representation of dozens of variants reflectingpositions that only diverge in some aspects
2 Many guidelines cannot be represented as logical sentences because theyrepresent lsquorecommendationsrsquo or vague requirements eg see (GC3) or (GC4)(Table 2)
These issues suggest that a combination of formal sentences with reified guidelinesprovide the best option for flexibility The representation of guidelines has beenstudied in the field of HumanndashComputer-Interaction (HCI) in the last decade(Vanderdonckt 1999) and the main philosophy of such representations remainsvalid in the learning design domain The major challenge for such representations isthat of providing computational semantics that could be used for the implementationof software support but this is difficult to attain in all the cases Thus the concept oflearning design guideline will provide two kinds of representations general andschema-specific
The general use of instances of oc_NormativeSpecification allows the model-ling of lsquoconceptual works that describe how something should bersquo This may includerepresentation of known normative representations or procedures For example onemay refer to a concrete learning design model as that derived from Bannister (1981)which states that creativity cycles follow an initial phase of brainstorming (idea cap-ture) followed by a phase in which key issues are focused into an operation strategy(idea development) and a third phase in which project control takes place through alsquorecursive cyclersquo (operational management) The reference could be made by pointing tothe academic reference describing it as a conceptual work or it could be described asan Activity-Spec modelled explicitly inside the ontology As another examplesome detailed guidelines could be modelled by the normative specification sub-category of oc-Instructions defined as lsquoa sequence of tasks to be performed byan agentrsquo
Schema-specific guidelines restrict the freedom to represent criteria as conceptual workinstances to a concrete and commonly agreed ontology that determine what can be saidabout how learning designs should be according to a theoretical standpoint or frame-work These build on an activityndashrole ontology as the one described in the previoussection and add a prescription on the kind of lsquoshould-bersquo predications that are allow-able Table 3 describes examples of a minimal language that could be used for thatpurpose
Semantic learning designs 345
copy 2006 The Author Journal compilation copy 2006 British Educational Communications and Technology Agency
In the examples CAT represents a category of individual expressed through a conceptor a logical formula and the sentence is a synthetic way of expressing some recom-mended or mandatory constraint over such category Other languages could be engi-neered in further research but the one described can be directly expressed in OpenCycand combined with software modules that check or enforce them in some way Notethat this representation does not use direct OpenCyc sentences so that reification is usedinstead of direct logical sentences It is not the aim of this paper to determine the mostpowerful and flexible representation language but only set a foundation for furtherspecification efforts
Analysis and comparison of two theoretical semantic frameworksAs an assessment of the representational capabilities of the scheme just described acase study of representation is sketched in this section We will use the learning designtoolkit described by Conole et al (2004) as the underlying terminology even thoughother frameworks could be used instead
Two rather diverging positions for design will be used as an illustration Table 4 sum-marises the two design scenarios
Table 3 Possible language for use in learning designs
Sentence type Potential uses Example
RecommendationshouldBe (CATSENT)
Recommend kinds ofactivities
Design activities are recommended (GC3)shouldBe(Method subEvent (CATDesignActivity))
ObligationmustBe (CATSENT)
Some activity typesmust follow others
Assessments must follow each expositionto contents (GD1) mustBe(and(ExpositiveActivityIndividualActivity)startAfterEndingOf(CAT
AssessmentActivity)
Activities must havesome characteristic
Every activity must entail interactionmustBe(LDActivity(usedIn(CAT
InteractiveService)))Every activity must provide acontent-oriented learning object forindividual use (similar to GB1)mustBe(LDActivity(and (usedIn(CAT
LearningObject)(learnerRoleNumber(1))))
Some structure shouldbe enforced in methods
Methods should have a sequential actstructuremustBe(Methodnot ((subEvent(CAT Play))))
CAT = category GC3 = Constructivist Guideline 3 (see Table 2) GD1 = Guideline 1
346 British Journal of Educational Technology Vol 37 No 3 2006
copy 2006 The Author Journal compilation copy 2006 British Educational Communications and Technology Agency
The following subsections sketch the main outcomes of an example learning designprocess for both scenarios assuming that the process described by Conole et al (2004)is used in both situations
Differences in modelling the design processThe forces driving the design process in both scenarios are fairly diverging The designof Scenario A will typically be constrained by a clear target specified in terms of anaggregated competency level while in Scenario B there are no concrete competenciesto be acquired and the main objective of the design activity would be twofold On theone hand the activities are targeted at promoting interchange of ideas in an attemptto stimulate casual convergence of interest which is the main value of the culture ofthe special interest group On the other hand the activities have the generic objectiveof strengthening social ties which is an important element of cohesion for the under-taking of joint projects
Table 5 describes the main inputs and possible activities and their objectives at a highlevel
The changes intended to be achieved in each of the scenarios are stated differently witha formal definition in (A) and a shallow cognitive and social representation in the case
Table 4 Learning design scenarios
Scenario DescriptionClassification according to
Conole et al (2004) toolkit
(A) Systemic-Behaviourist A software development company attempts to designlearning units about newtechnologies or softwareproducts for its developerstaff The design is drivenby a cycle of lsquocompetencygap analysisrsquo that modelsorganisational learning in terms of the aggregated competencies of the employeesmdashsimilar to theone described by Sicilia(2005)
Focuses on the individual where a body of informationforms the basis and rawmaterial for learning Learning is mainly considered as nonreflective ie based on skill learning and memorisation
(B) Sociocultural A special interest group ona concrete research topicaims at developing the knowledge of its membersby sharing experiencesand discussing on themesof common interest
Focuses on the social whereexperience is the main source of learning Conscious reflection on experience is transformed into learning
Semantic learning designs 347
copy 2006 The Author Journal compilation copy 2006 British Educational Communications and Technology Agency
of (B) In addition case (B) requires a model of social network that is largely irrelevantin case (A)
Differences in recording activity rationalesThe approach for the recording of design rationales also differs because of the differentcontexts Table 6 summarises the main elements for each of the scenarios
Conclusions and directions for researchCurrent technology and practice related to the description of reusable digital learningresources has resulted in techniques that allow the description of learning activities of
Table 5 Inputs activities and objectives of two learning design scenarios
Scenario Design inputs Activities and their objectives
(A) Systemic-Behaviourist A formal competency recordfor employees and competencyrequirements expressed in terms of amounts of required competencies (assuming a shared measurement scale) Such requirements are actually derived from strategy or from the contingent requirements of upcoming projects
Competency instances are used to measure the levelof Competency-Spec for the required competencies (or their constituents) A formal match considering the agenda of employees is carried out for determining the target learners LearningObjects forindividual use are selected asthe main elements of theactivities and the criteria forselection are again a match ofthe objectives stated inmetadata (as stated incategory 9 of Learning Object MetadatamdashClassification) and thecompetencies required
(B) Sociocultural The expertise of the special interest group members as stated in the concepts addressed by their publicationsprojects or other activities
Informal group activities are the main approach applicablesince there is not a predefinedgroup structure but a cultureof sharing and socialisation The results of the sharing of ideas can be expressed in terms of oc-knowsAboutpredicates and social ties will be modelled through positive valence FOAF-knows(httpxmlnscomfoaf01)predicates with an added numerical value representingstrength
348 British Journal of Educational Technology Vol 37 No 3 2006
copy 2006 The Author Journal compilation copy 2006 British Educational Communications and Technology Agency
a complex kind including their intended participants educational objectives and exe-cution constraints The representation and recording of the assumptions and guidelinesthat were considered in the creation of such resources would extend current techniquesto capture the rationale of the design itself This in turn will eventually result in theavailability of a large base of empirical data available in which theoretical assumptionsor rules are explicitly connected to the resulting designs as these could be linked withthe actual results and trace of the learning processes that used them The resulting bodyof evidence would represent a critical asset in the development of the theory of learningmediated through computers
This paper has described the first attempt to extend existing learning technology modelswith a schema explicitly targeted at describing learning design rationales The OpenCycknowledge base has been used as the semantic reference from which the ontologicalschema has been specified Concretely an activity-based schema based on IMS LD hasbeen described and theoretical assumptions and codified guidelines have been intro-duced The flexibility of the representation schema has been analysed through therepresentation of two stereotyped theoretical positions
Two main research directions should continue the work described here On the onehand further ontological engineering effort is required to come up with a more com-prehensive and rich schema for learning design rationales On the other hand tool
Table 6 Main elements of the two learning design scenarios
Scenario Approach Examples
(A) Systemic-Behaviourist Formal logical sentencesare used to guarantee that every design followsan organisationalobjective
It is required that the objective of every Method is attached to a Competency-Spec instance as objective
Formal assessment is a requirement foraccountability purposes
Guideline (GD1) is attached to everyDesigningLearningResource instance that produces an Act
(B) Sociocultural The interaction in informal settings can beused to derive social models (Sicilia amp Garciacutea2004)
The interactions in discussion canbe used to measure the degree of lsquoproximityrsquo between members Therationale of the activities can bejustified in terms of SocialTies inside SocialNetworks Further the focus of each groupactivity should be clearly stated anda combination of the increase in collective knowledge and the strengthening of social relations can be used for assessment
Semantic learning designs 349
copy 2006 The Author Journal compilation copy 2006 British Educational Communications and Technology Agency
implementation and record gathering efforts would be required to attain the ultimateobjective of the ontological definitions described so far With such effort an unprece-dented research material could be available in some years providing improved oppor-tunities for scientific inquiry on design as an activity regarding pedagogy
Other potential directions for research may be the description of interpretations of thedesign intentions (for which no rationale was explicitly recorded) the mining or detec-tion of patterns as related to rationale description or the detailed use of theoreticalrationales in agents that support the learning process which represent a category ofrelevant learning support software (Holmes in press)
ReferencesAllert H (2004) Coherent social systems for learningmdashan approach for contextualized and
community-centred metadata Journal of Interactive Media in Education 2004 1Baader F Calvanese D McGuinness D Nardi D amp Patel-Schneider P (Eds) (2003) The
description logic handbook Theory implementation and applications Cambridge CambridgeUniversity Press
Bannister D (1981) Personal construct theory and research method In P Reason amp J Rowan(Eds) Human inquiry a sourcebook of new paradigm research (pp 191ndash200) Chichester UKJohn Wiley and Sons
Berners-Lee T Hendler J amp Lassila O (2001) The semantic web Scientific American 284 534ndash43
Brown A amp Voltz B (2005) Elements of effective e-learning design International Review ofResearch in Open and Distance Learning 6 1
Buzza D Bean D Harrigan K amp Carey T (2005) Learning design repositories adaptinglearning design specifications for shared instructional knowledge Canadian Journal of Learningand Technology 30 3
Conole G Dyke M Oliver M amp Seale J (2004) Mapping pedagogy and tools for effectivelearning design Computers amp Education 43 1ndash2 17ndash33
Dodero J M Aedo I amp Diacuteaz P (2002) Participative knowledge production of learning objectsfor e-books The Electronic Library 20 4 296ndash305
Friesen N (2005) Interoperability and learning objects an overview of e-learning standardiza-tion Interdisciplinary Journal of Knowledge and Learning Objects 1 1 23ndash31
Gruber T (1995) Towards principles for the design of ontologies used for knowledge sharingInternational Journal of HumanndashComputer Studies 43 56 907ndash928
Hatchuel A (2002) Towards design theory and expandable rationality the unfinished programof Herbert Simon Journal of Management and Governance 5 3ndash4 260ndash271
Holmes J (in press) Designing agents to support learning by explaining Computers and EducationIEEE Learning Technology Standards Committee (2002) Learning object metadata (LOM) Final
Draft Standard IEEE 1484121ndash2002IMS (2003) IMS Learning Design Information Model (Version 10) Final Specification 20
January 2003Kaptelinin V Kuutti K amp Bannon L (1995) Activity theory basic concepts and applications
In Blumenthalet al (Eds) Humanndashcomputer interaction Lecture Notes in Computer Science 1015189ndash201
Koper R (2004) Use of the semantic web to solve some basic problems in education increaseflexible distributed lifelong learning decrease teacherrsquos workload Journal of Interactive Mediain Education 2004 6
Lenat D (1995) CYC a large-scale investment in knowledge infrastructure Communications ofthe ACM 38 11 33ndash38
350 British Journal of Educational Technology Vol 37 No 3 2006
copy 2006 The Author Journal compilation copy 2006 British Educational Communications and Technology Agency
McGreal R (2004) Learning objects a practical definition International Journal of InstructionalTechnology and Distance Learning 1 9 21ndash32
Packer M amp Goicoechea J (2000) Sociocultural and constructivist theories of learningontology not just epistemology Educational Psychologist 35 4 227ndash241
Palmer J (2001) Fifty major thinkers on education London and New York RoutledgePolsani P R (2003) Use and abuse of reusable learning objects Journal of Digital Information 3
4 Rothwell W amp Kazanas H (1992) Mastering the instructional design process San Francisco CA
Jossey-BassSaacutenchez-Alonso S amp Sicilia M A (2004) Relationships and commitments in learning object
metadata In Proceedings of the 5th International Conference on Information Technology BasedHigher Education and Training ITHET 2004 Istanbul Turkey
Sicilia M A (2005) Ontology-Based Competency Management Infrastructures for theKnowledge-intensive Learning Organization In Lytras amp Naeve (Eds) Intelligent Learning Infra-structures in Knowledge Intensive Organizations A Semantic Web perspective IDEA USA pp 302ndash324
Sicilia M A amp Garciacutea E (2004) Fuzzy group models for adaptation in cooperative informationretrieval contexts Lecture Notes in Computer Science 2932 324ndash334
Sicilia M A amp Garciacutea E (2005) On the convergence of formal ontologies and standardizede-learning Journal of Distance Education Technologies 3 2 13ndash29
Sicilia M A amp Lytras M (2005) On the representation of change according to different ontol-ogies of learning International Journal of Learning and Change 1 1 66ndash79
Sicilia M A Lytras M Rodriacuteguez E amp Garciacutea E (in press) Integrating descriptions of knowl-edge management learning activities into large ontological structures a case study Data andKnowledge Engineering
Tennant M (1991) Expertise as a dimension of adult development New Education 13 1 49ndash55Vanderdonckt J (1999) Development milestones towards a tool for working with guidelines
Interacting with Computers 12 2 81ndash118Wand Y amp Weber R (1995) On the deep structure of information systems Information Systems
Journal 5 203ndash223Wiley D A (2001) The instructional use of learning objects Bloomington IN Association for
Educational Communications and Technology
332
British Journal of Educational Technology Vol 37 No 3 2006
copy 2006 The Author Journal compilation copy 2006 British Educational Communications and Technology Agency
and Conole Dyke Oliver and Seale (2004) have provided a model for designing e-learning activities from different theoretical positions In addition learning technologystandards that model learning contents and activities (Friesen 2005) provide enhancedopportunities for the reuse of diverse kinds of resources stemming from the concept oflearning object (Polsani 2003 Wiley 2001) These resources range from self-standingeducational materials to complex learning activity designs encompassing the interac-tion of several roles and being driven by rules explicitly provided to guide the flow ofactivities
Concretely the increasing adoption of the
IMS Learning Design
(LD) specification(httpwwwimsprojectorg) has provided a standardised common ground to therepresentation of learning designs in digital form An
activity
in IMS LD is consideredas a piece of interaction among a number of specified rolesmdashplayed by persons ininstantiations of each concrete learning designmdashthat produce a tangible outcome byusing a concrete environment The environment is in turn made up of
learning objects
(McGreal 2004) and
services
(ie facilities available at runtime as for example a chatroom) Activities can be further decomposed in subactivities and they are aggregatedinto
methods
that specify the conditions for application of the learning design alongwith the planned objectives that will eventually match the outcomes of the activitiesMethods can be structured around concurrent
plays
and these in turn can be structuredin sequential
acts
the latter allowing the specification of conditional execution Thisschematic description of learning design gives an idea of the flexibility the specificationprovides in describing activity-based learning programs as acknowledged elsewhere(Allert 2004) IMS LD combined with learning object-oriented specifications suchas IEEE Learning Object Metadata (LOM) (IEEE 2002) and ADL-SCORM (httpwwwadlnetorg) enable a degree of representation of activities and materials withunprecedented applications
As can be appreciated from the earlier discussion there exists a growing consensus ona concrete terminology and model for creating metadata describing learning activitiesand their associated constituents and at least general principles for their design are yetavailable Nonetheless while this is enough for describing the activities by themselvesa richer framework is required to capture also the intellectual process that was involvedin their crafting ie the hypotheses assumptions and decisions that led to the concretearrangement of activities and conditions Doing so will expand the current scope ofmetadata-described learning resources from the description of the activities themselvesto the description of the designerrsquos intentions It may be argued that such kind ofknowledge is difficult to capture because of the heterogeneity of theoretical standpointsand also because of the use of lsquotacitrsquo knowledge or little reflexive designs that could befound in many designs But the benefits available for researchers and practitioners ingeneral are still important enough to deserve the effort and in fact toolkits as the onedescribed by Conole
et al
(2004) consider design theories explicitly and can be used asa point of departure The benefits of recording the rationale for learning designs iscomplementary to the
postactivity
analysis of the actual history and outcomes of activ-ities in actual settings which has been pointed out as a relevant source of empiricalinformation (Koper 2004) that will eventually be available in the coming years
Semantic learning designs
333
copy 2006 The Author Journal compilation copy 2006 British Educational Communications and Technology Agency
More concretely the modelling of the rationale of learning designs would provide anumber of benefits for researchers and educators in general Among these benefitsSicilia and Lytras (2005) provided the following list (1) the
linking of theoretical assump-tions to practical learning designs
for informative purposes (2) the use of such
links asresources for education of learning designers
(3) the search for
patterns in design situations
that include some given theoretical assumptions and (4) the lsquoinferencersquo of
detailed com-parisons between the effectiveness and adequacy of learning designs
(for similar settings)These aspects are reflections of the practical benefits of the main objective of the record-ing of design rationales which is no other than advancing the scientific inquiry onlearning by enabling the recording of explicit links from design rules to concretedesigns and the latter to the outcomes of the lsquoinstantiationrsquo of these designs in possiblydiverse institutions or groups This would come up with an unprecedented empiricalbase for meta-analysis and knowledge discovery For example hypotheses on the ade-quacy of group-based or individual learning for some given conditions could be testedwith a large empirical base containing the history of interaction and results of multiplesimilar learning programs
This paper approaches the problem of the metadata representation of assumptions andhypothesis related to learning designs in a general sense Because the entities to berepresented are relevant rules or rationales involved in the intellectual process of design-ing learning experiences the schema described is deliberately open-ended This is arequirement imposed by the fact that theories and guidelines related to learning arebased on models that are abstract in nature This in turn is because they attempt tocharacterise learning as a general human cognitive process but there exist at least somerecurring elements that are common to many practical learning designs We have notattempted to build an ontology of existing theories of learning but addressing somelsquoupperrsquo elements that cross-cut any of them eg notions of
change
are one of these upperelements (Sicilia amp Lytras 2005) In any case the fact that diverging accounts oflearning exist is far from controversial and even modern theories somewhat diverge intheir basic ontological commitments (Packer amp Goicoechea 2000)
The representation described departs from a model based on IMS LD but expressed in aricher ontological language Ontology engineering (Gruber 1995) provides a methodto develop semantic conceptualisations and
Semantic Web
technology (Berners-LeeHendler amp Lassila 2001) aims at providing tools and techniques to develop softwarethat exploits them in the context of a learning system Formal ontology provides theknowledge representation infrastructure for ontologies of learning to a level of consid-erable complexity and richness as supported by description logics (Baader CalvaneseMcGuinness Nardi amp Patel-Schneider 2003) It should be noted that the lsquoontology oflearningrsquo explicitly or tacitly assumed actually constraints and drives the behaviourof the system as a whole and the criteria used for the evaluation of the outcomes ofactivities also become a reflection of previous ontological commitments For examplean ontology not considering social issues will not allow (or will not care about) thetracking of the evolution that takes place in the social network of the learners involvedas the activities are carried out
334
British Journal of Educational Technology Vol 37 No 3 2006
copy 2006 The Author Journal compilation copy 2006 British Educational Communications and Technology Agency
The resulting learning design descriptions are called here lsquo
semantic learning design
rsquo torepresent the fact that they are described in ontological terms and that the meaning ofthe design as conceived by the designer is explicitly represented This carries one stepfurther the current metadata descriptions provided by specifications such as LOMSCORM or LD
As a technique for validating the semantic precision of the schema described hereexplicit links have been provided to concepts and relations that are already described ina large upper ontology concretely the
OpenCyc
09 knowledge base This is an alterna-tive to analysis techniques such as the BungendashWandndashWeber (Wand amp Weber 1995)that fosters the reuse of existing open knowledge engineering The subsequent mappingto modern Web-enabled ontology languages such as OWL Web Ontology Language(httpwwww3orgTRowl-ref) is a straightforward step In addition some represen-tative examples of different theoretical standpoints are described as illustrations
The rest of this paper is structured as follows Section 2 provides an ontological accountof an activity-based model of learning activities as supported by IMS LD which consti-tutes the
artefact
that was created by an intellectual process Then Section 3 describesa proposal for an ontological schema targeted at recording design rationales in a broadsense The scheme described is then used in the fourth section to provide as a case studythe comparison of two theoretical frameworks that lead to two different descriptions fora concrete learning design Finally conclusions and future research directions are pro-vided in the fifth section
Recording educational metadata and semantic annotation of learning activities
The first step in the modelling of learning design rationales is that of representing theresults or
artefacts
of the design process In this section the main elements of the IMSLD schema are expressed in ontological terms considering both the design in itself (iethe equivalent to a learning design description) and the entities that result from theapplication of designs in actual occurrences of learning following them (ie the record-ing of actual executions of the planned design in concrete instructional or trainingsettings)
A significant amount of reuse in terminological structures can be achieved by buildingconceptualisations on top of existing large terminological bases like
OpenCyc
(httpwwwopencycorg)
OpenCyc
is the open source version of the
Cyc
knowledge base(Lenat 1995) which contains over 100 000 atomic terms and is provided with anassociated efficient inference engine
Cyc
uses as its underlying definition language avariant of predicate calculus called
CycL
and it attempts to provide a comprehensiveupper ontology of lsquocommon sensersquo knowledge In the rest of the paper the ontologicaldefinitions provided are explicitly linked to
OpenCyc
definitions (in which ontologicaldefinitions of terms and relations are provided in courier font and those that are definedin
OpenCyc
are prefixed by lsquoocrsquo) as a means of reusing existing ontological engineeringefforts and also as a way of validating the concepts being represented
Semantic learning designs
335
copy 2006 The Author Journal compilation copy 2006 British Educational Communications and Technology Agency
Describing the main structural elements of IMS LD
The conceptual model of IMS LD (IMS 2003) provides a concise lsquometalanguagersquo thatcould capture a diversity of pedagogical approaches In doing so it essentially adheresto an activity-centred paradigm (Allert 2004) It should be noted that this model isneutral to specific learning theories Even though it integrates the hierarchical struc-ture of activities and mediation principles of
Activity Theory
(Kaptelinin Kuutti amp Ban-non 1995) it is actually a metalanguage that does not make assumptions about otherelements of such theory as the internalisationexternalisation differentiation
The first step in developing an ontological schema for IMS LD must be that of clarifyingthe main entities to be represented
OpenCyc
provides a rich framework to describetemporal activities concretely the
oc_PurposefulAction
concept (a kind of
oc_Event
) defined as lsquoan action consciously volitionally and purposefully done by atleast one actorrsquo captures the essence of learning activities which are the ultimate targetof the specification Nonetheless IMS LD metadata does not attempt to representing theactivities as actually enacted by persons but they describe
blueprints
that will eventuallybe bound to one or several instantiations of activities and they are intended to besearched as such (Buzza Bean Harrigan amp Carey 2005) In contrast
oc_Event
s andtheir associated entities are related to instantiations and not to their specificationsFurther the concept of
oc_ActorSlot
represents the specific roles played by personsin events so that this could be used to represent the LD
staff
and
learner
role categorybut they do not represent roles understood as templates for their eventual binding toconcrete people
An alternative to representing LD specifications as blueprints is that of creating
Event
types for each activity defined in a learning design But this approach entails at leasttwo problems On the one hand this is not ontologically correct because it does notrepresent the reality that the production of IMS LD is an act of creating specificationartefacts instead of a process of classification of possible real world entities In factmany learning designs will probably be never executed and some of them may bepurposefully created as mere design examples On the other hand the direct creationof learning designs as ontology terms entails that the activities actually performed mustadhere strictly to the original learning design used which severely restricts the flexibil-ity in using the design with ad hoc changes or modifications (or even modified
versions
)This second practical flaw suggests that LD specifications should be represented asseparated entities
This has led us to the major decision of modelling learning designs as specificationsfollowing the definition of
oc_Specification
as lsquoan abstract work that constitutesa description of the properties of a
Situation
or a
SomethingExisting
and sometimes evenentire collections of such thingsrsquo Table 1 summarises the representation of
LearningDesigns
in terms of specifications (specialisations of
oc_Specification
) their mainrelated properties and the implications of each of the specifications when producingactual activities Notifications and completion conditions are omitted for brevityThe concepts of lsquo
learning design
rsquo and lsquo
method
rsquo can be considered as equivalent from an
336
British Journal of Educational Technology Vol 37 No 3 2006
copy 2006 The Author Journal compilation copy 2006 British Educational Communications and Technology Agency
Table 1 Summary of the main specification definitions in the ontological schema
Concept Main descriptions Consequences in actual activities
Method-Spec Objectives and prerequisites(described in the followingsection) Structured in
play
specifications or directly
act
specifications
Instantiation is an independentmeaningful learning activity Playsare cotemporal
oc_subEvent
ssubsumed in the method Note thatthe semantic definition of subeventsyet entails the subsuming of subeventsin the containing event
Play-Spec Collection of sequential acts thatcomprise the play
Because plays are potentiallyconcurrent there are no furthertemporal restrictions
Act-Spec Collection of concurrent role parts that detail the activitiesincluded in the design Role partsand environments are the mainelements related to acts
The sequence of instantiations of actscould be sequenced through the
oc_startsAfterEndingOf
RolendashPart-Spec The linking of role specificationsto concrete activities or activitystructures
This entity is not required in theontological schema because theparticipation of roles in the activitiescan be stated in terms of specificationsof
oc_ActorRole
s or subeventsthat are not of a temporal nature
Activity-Spec The specification of learning andsupport structures in terms of
Role-Spec
s
ibtUsed LearningObject
s and used
Service
s Concrete objectivescan also be attached
Each instantiation of an
Activity-Spec
will result ininstances of specific types of
oc_Activity
Following LD specs
support
activities instantiate for everyrole mapping while
learning
onesinstantiate for all the role mappings
Environment These are descriptions of learning objects and services that do not require anexplicit mapping in the ontological schema
Service Service facilities have to belsquoinstantiated by a local run-timeservicesrsquo eg a chat or discussionforum Their computationalnature can be described interms of concrete
oc_ComputerProgramCW
Subcategories of such conceptualrepresentations of softwareentities can be used for thetypology of services
The
oc_ComputerProcessRunning
term is used to describe executionsof
oc_ComputerProgramCW
which cover every instance of aservice as understoodin IMS LD Thelinking between conceptual andactual running instances is coveredby the required predicate
oc-programOfProcess
Semantic learning designs
337
copy 2006 The Author Journal compilation copy 2006 British Educational Communications and Technology Agency
Learning object Those descriptions provided bylearning object metadataLearning objects can be definedas
oc_IBT
s as described inSicilia Lytras Rodriacuteguez andGarciacutea (in press) and approachesto describing metadata throughontologies yet existmdashsee an overview in Sicilia and Garciacutea(2005)
The learning objects selected arerepresented as instances in theontology which are declared to beused in activities through the
ibtUsed
predicate
Role-Spec Describe stereotypical actors oflearning and support activitiesThese act as lsquocontractrsquo for theactual participants in the senseprovided by Saacutenchez-Alonso andSicilia (2004)
Role specs are mapped to actualusers when the activities take placeThis is mapped to predicates of theclass
oc_ActorRole
Concept Main descriptions Consequences in actual activities
LD
=
Learning Design
Table 1 Continued
ontological perspective as the former allow onemdashand only onemdashinstance of the latterwith no particular logical differences
As described in Table 1 the
OpenCyc
built-in support for modelling temporal events canbe used to express the coordination relationships between the constituents of the unitsof learning The actual instances of learning events will be modelled as instances of
oc_Learning
(lsquothe collection of all events brief or extended in which an agent isacquiring information or know-howrsquo) and more specifically as instances of a speciali-sation
ComputerMediatedLearning
simply requiring the use of information net-works for at least part of the learning activities Because events in
OpenCyc
are alsosituations they have components in addition to the temporal extent including peopleinvolved information things used and other arrangements
Figure 1b depicts the main elements that describe actual current or past enactmentsof activities The schema is based on a clear separation of the specifications and theactual activities that by itself constitutes a metadata layering
OpenCyc
predicatesas
oc_situationConformsToSpecification
allows the linking of situations(every action is a situation also) to the specifications they conformmdashbe it purposely ornot The
LD-Activity term has been introduced as the common category of learningand support activities Methods plays and acts become special subcategories of LD-Activity and the constituent activities of Acts are instances of LD Activity also Rolesare attached to any kind of LD Activity even though it is required for the low-levelstructure The learning design concept of lsquoactivity structurersquo as an aggregation ofactivities does not require an explicit modelling because of the generic mereology ofactivities engineered in OpenCyc
338 British Journal of Educational Technology Vol 37 No 3 2006
copy 2006 The Author Journal compilation copy 2006 British Educational Communications and Technology Agency
Figure 1 (a) Summary of the competency components in the ontological schema (b) Summary of the main activityndashrole definitions in the ontological schema
oc_WorkingEvent
oc_IntelligentAgent
The competency elements aremeasurable characteristics andthey will be typed by the relatedSpecification classes eg Skill
SpecputlnPracticeln
(a)
Competency Competency-Spec
Competencyelement
Skill oc_ Competence
Attitude
Knowledge
has
thingSpecified
oc_Learning
ComputerMediatedLearning
oc_Role
(b)
Staff
Learner
oc_Specification
Method-Spec
Play-Spec
Act-Spec
Support
LD-Activity
contains
contains
Method
Play
Act
startAfterEndingOf
subAction
subAction
Activity-Spec
ltltActorSlotsgtgt
situationConformsToSpec
Activity-Spec
Activity
Semantic learning designs 339
copy 2006 The Author Journal compilation copy 2006 British Educational Communications and Technology Agency
The concept of oc_Role accounts not only for participant roles but also for otherkinds of lsquocomponentsrsquo This is why two specialisations of the concept ofoc_ActorSlot labelled Staff and Learner are required to differentiate partici-pants from other elements As is obvious from the earlier discussion the built-in rep-resentation of activities in OpenCyc is powerful enough for the definition of learningdesign activities
The final part of the ontological specification described so far is that of the mappingfrom LD specifications to the actual past or present activity realisations Theoc_situationConformsToSpecification predicate can be used to trace whichactivity realizations come from a given learning design activity definition and the sameoccurs with the oc_thingSpecified for other learning design elements such asroles or environments Existing learning design execution engines as CopperCore(httpcoppercoreorg) actually map LD specifications to online collaboration so thatthe translation of the ontological representations of learning designs to ontology defi-nitions of activities is straightforward following similar representation
Describing objectives outcomes and preconditionsAccording to IMS learning design learning objectives and prerequisites lsquocan bedescribed using the IMS Reusable Definition of Competency or Educational Objective(RDCEO) format but can also refer to simple resources (eg a text) with a descriptionof the learning objectiversquo As the latter option is not useful from the viewpoint of com-putational semantics we will adhere to the former As commented earlier the compe-tency view is not exclusive to other kinds of objectives related to other ontologicalstandpoints (Sicilia amp Lytras 2005) and other conceptions of change or improvementmay include social ties cognitive structures or the combination of the knowledge of agroup In fact competencies adhere mostly to behaviouristic frameworks that focus onobservable external behaviour (Tennant 1991) rather than on cognitive representa-tions Nonetheless the further discussion is equally applicable to other kinds of objec-tives so that we will adhere to the competency view
oc_Competence in OpenCyc is defined as a quantity that lsquofor describing the level ofskill with which an agent performs some taskrsquo This does not remove the lack of preci-sion that is yet present in RDCEO definitions as described by Sicilia (2005) As analternative Figure 1a provides a simplified account of the model described by Sicilia asrelated to OpenCyc terms in which competence is integrated as one of the elements of thebroader notion of competency according to the model of Rummel (Rothwell amp Kazanas1992) which considers work situations attributes of the individual that are applied insuch situations and measurement with respect to some standard of performance Themost relevant distinction is again the separation of actual competency representationsand competency lsquodefinitionsrsquo (ie specifications) intended to provide a type structure tocompetencies to be used in objective or precondition specification In this case thegeneric oc-thingSpecified predicate can be used for the linking of both levels ofrepresentation
340 British Journal of Educational Technology Vol 37 No 3 2006
copy 2006 The Author Journal compilation copy 2006 British Educational Communications and Technology Agency
Objectives and preconditions can then be expressed as formulas on the acquisition ofcompetencies or their components (competency elements) Descriptions specific toOpenCyc may use oc_CycLSentence-Assertible but simpler enumeration ofcompetency components could be used instead
Describing learning design rationales theories hypotheses guidelines and rulesThe concept of lsquodesignrsquo for our purposes is properly captured in the 5a definitionprovided in the Merriam Webster Online Dictionary (wwwm-wcom) lsquoan underlyingscheme that governs functioning developing or unfoldingrsquo In consequence learningdesigns refer to artefacts and not to the process of devising the educational solutionThis is equivalent to the concept oc_Design subsumed by the notion ofoc_Specification which is described as lsquoan abstract work that constitutes adescription of the properties of a Situation or a SomethingExisting and sometimeseven entire collections of such thingsrsquo In consequence the term LearningDesignwould simply delineate a subset of specifications that are oriented to describe howlearning experiences with concrete objectives should be Trivially oc_Designerdescribes the authors of the learning designs and oc_Designing represent theoc_HumanActivity that leads to the production of learning designs Designing as aspecial kind of activity is carried out (oc_doneBy) by a designer which provides a wayto attach metainformation about the rationale of the design
From an ontological perspective the departing problem in modelling learning designrationales is that their assumptions and hypotheses of different theoretical standpointsneed not be compatible or need not consider the same aspects of reality (Conole et al2004 Packer amp Goicoechea 2000) Furthermore different pedagogical approaches areknown to result in different mappings to activities (Conole et al 2004) This wouldeither require the provision of separate ontologies or the use of a representationalmechanism that allows such kind of potential inconsistency or divergence The conceptof microtheory in OpenCyc provides such representational mechanism intended to orga-nise assertions that depend on lsquoshared set of assumptions on which the truth of theassertions dependsrsquo Definitions inside the same microtheory need to be consistent butthis is not required across microtheories The descriptions provided in the earlier sectioncan be used as a base microtheory called LearningActivitiesMt for others that are spe-cific to concrete theoretical frameworks In this section we will describe elements thatare common to the description of theoretical frameworks the distinction among micro-theories will not be discussed
The intellectual processes of learning design in themselves will be represented asoc_Designing activities carried out by oc_Designers An additional consider-ation is that the information about the rationale of the design would be better attachedto representations of each of the design processes (ie to oc_Designing instances)than to the final design This enables the registering of different assumptions for differ-ent contributions to the same artefact eg different revisions from the same of differentdesigners
Semantic learning designs 341
copy 2006 The Author Journal compilation copy 2006 British Educational Communications and Technology Agency
Several designers may participate in the same designing activity and formal frameworksfor collaborative learning resource creation (Dodero Aedo amp Diacuteaz 2002) can be inte-grated by clearly specifying which class of things can be the outcome of a designingprocess For the sake of flexibility the concept of LearningDesignPart is introduced as anall-encompassing category that includes each of the elements described in the previoussection This includes not only activities and roles but also objectives conditions andthe like Furthermore the granularity of the descriptions allows for different levels fromconsidering a unique designing activity for the whole package to recording each ele-ment of the learning design comprehensively Because oc_Actions are subsumed bythe concept of oc_Event considered as a lsquodynamic situation in which the state of theworld changesrsquo each edition of a metadata record or any other form of description maybe subject to be described The concept is flexible enough to be able to record fine-grained detailed decisions This of course may include different versions of the sameartefacts but this is an orthogonal aspect of configuration that is not relevant to ourpresent discussion
Figure 2 depicts the main elements of the loci to which rationales assumptions andguidelines will be attached
A LearningDesigner can be defined as the class of humans that performactivities of learning resource design The consideration of learning design as aunique human activity could be considered as restrictive of intelligent softwarefor the task Simply changing the oc_Designer class to a subconcept ofoc_IntelligentAgent could remove this but this entails a change in Open-Cyc knowledge base The concept oc_Designing is a subclass of the conceptoc_ThinkingAboutTheStateOfTheWorld which represents lsquothinking pro-cessesrsquo which include mental events such as planning evaluating or inventingThus the concept of design described so far should be understood in terms of the
Figure 2 Main elements of the ontological schema for the loci of design rationales
oc_performedBy modifies
342 British Journal of Educational Technology Vol 37 No 3 2006
copy 2006 The Author Journal compilation copy 2006 British Educational Communications and Technology Agency
concept of lsquoexpanded rationalityrsquo as described by Hatchuel (2002) integrating cre-ativity and unexpected expansions of the original requirements This precludes onto-logical definitions in which the problem space is completely bounded a priori Inconsequence a degree of openness is necessary to integrate different kinds of detailin description from fully described ones to others with shallower semantics eg someproviding only references to generic assumptions
Theoretical frameworksWe will make the distinction that guidelines (as further defined) are the propositions thatdirectly predicate about the validity or appropriateness of learning design parts whileother theoretical propositions about learning that are not directly saying anythingabout the design of digital learning resources are considered as assumptions followingthe general sense provided in the Merriam-Webster dictionary lsquoa fact or statement (asa proposition axiom postulate or notion) taken for grantedrsquo
The fact that some assumption hypotheses or proposed law about learning is sustainedby a broader empirical base or is better sustained by rational means is out of the scopeof this work because here we focus on design practice (even though it is of an obviousinterest for researchers) In principle even unorthodox positions may be sustainedbecause the mechanism of microtheories and their inclusion mechanism allows forgreat flexibility in modelling them
Table 2 provides some examples of theoretical assumptions and related guidelines forlsquoextremersquo archetypical standpoints on learning These are not intended to reflect actualcommon positions but they are provided as illustrative examples of radical positionsthat could be modelled with our ontological schema
The examples provided in Table 2 go a step further than toolkits in Conole et al (2004)in making the design rationale explicit Nonetheless in many cases lsquocommon sensersquo isapplied to design and a shallower approach to specification could be used instead egproviding simply links to the theoretical standpoints used It should be noted also thatthe resulting guidelines vary greatly in their level of concreteness and some of themmay be considered as mandatory from some standpoints while others are mere generalrepresentations This would lead to different computational strategies
Guidelines and rulesThe WordNet (httpwordnetprincetonedu) thesaurus defines guidelines as lsquoguidancerelative to setting standards or determining a course of actionrsquo or lsquoa rule or principlethat provides guidance to appropriate behaviorrsquo In our case guidelines are statementsthat serve as criteria for decision in designing learning Given that e-learning designis still far from being a mature discipline guidelinesmdashas instances of recordedknowledgemdashcan be characterised as oc_NormativeSpecifications becausethey describe how learning resource design activities and their outcomes lsquoshould bersquoin a propositional form Propositions lsquohave some truth value in some context or
Semantic learning designs 343
copy 2006 The Author Journal compilation copy 2006 British Educational Communications and Technology Agency
ldquoworldrdquo rsquo Furthermore propositions lsquoare assumed to be representable (at least in prin-ciple) by a sentence in some formal or natural languagersquo This raises the need for alanguage that expresses guidelines in a general sense There are two main options forsuch representation be them formal logical sentences or reified guidelines
Formal logical sentence representation entails the formalisation through sentences ofthe constraints of the designs For example if lsquoevery learning resource must be activersquoaccording to some theoretical standpoint (eg one that fosters reflective learning) thissentence may be represented as an axiom and put into the appropriate separate micro-theory as the following CycL fragment
($forAll X ($implies
($isa X $LearningObject)
($resourceType X $Active))
Then resources that are not active are simply excluded from the notion itself This is astrong position in the sense that precludes some views on the learning design process
Table 2 Some examples of theoretical assumptions and guidelines
Archetypicalstandpoint Assumptions Guidelines
Instructivist(Behaviouristic)
bull Learning occurs through association and reinforcement
bull The results of learning can be appreciated through observable outcomes
(GB1) Every instance of oc-Learning entails theexposure to some LearningObjectthat produces the association (GB2) Learning Activitysequences should provide assessmentactivities that confirm theassociations that should have beenestablished so far
Constructivist bull Learning proceeds by building mental structures
(GC1) Agentrsquos mental structuresshould be represented(GC2) These structures should bepre-assessed in order to becomparable with the resulting ones(GC3) Design Activities arerecommended(GC4) Contrast Activities arerecommended
Sociocultural bull Learning occurs through social participatio
(GSC1) Every instance ofoc-Learning entails theinteraction of severaloc-IntelligentAgents asoc-Actors
344 British Journal of Educational Technology Vol 37 No 3 2006
copy 2006 The Author Journal compilation copy 2006 British Educational Communications and Technology Agency
Another typical example may be that of a sociocultural guideline that preclude Methodsthat do not entail multirole activities that provide access to social interaction ServicesThis approach to guidelines provides the best framework for automation and wouldenable the creation of microtheories that represent clearly formulated views onlearning Nonetheless there are two major problems for the use of this option as theunique mechanism for encoding design rationales
1 There is no clear-cut distinction between models and theories of learning becausethey come from a history of schools of thought with interrelations (Palmer 2001)This would lead to the need for the representation of dozens of variants reflectingpositions that only diverge in some aspects
2 Many guidelines cannot be represented as logical sentences because theyrepresent lsquorecommendationsrsquo or vague requirements eg see (GC3) or (GC4)(Table 2)
These issues suggest that a combination of formal sentences with reified guidelinesprovide the best option for flexibility The representation of guidelines has beenstudied in the field of HumanndashComputer-Interaction (HCI) in the last decade(Vanderdonckt 1999) and the main philosophy of such representations remainsvalid in the learning design domain The major challenge for such representations isthat of providing computational semantics that could be used for the implementationof software support but this is difficult to attain in all the cases Thus the concept oflearning design guideline will provide two kinds of representations general andschema-specific
The general use of instances of oc_NormativeSpecification allows the model-ling of lsquoconceptual works that describe how something should bersquo This may includerepresentation of known normative representations or procedures For example onemay refer to a concrete learning design model as that derived from Bannister (1981)which states that creativity cycles follow an initial phase of brainstorming (idea cap-ture) followed by a phase in which key issues are focused into an operation strategy(idea development) and a third phase in which project control takes place through alsquorecursive cyclersquo (operational management) The reference could be made by pointing tothe academic reference describing it as a conceptual work or it could be described asan Activity-Spec modelled explicitly inside the ontology As another examplesome detailed guidelines could be modelled by the normative specification sub-category of oc-Instructions defined as lsquoa sequence of tasks to be performed byan agentrsquo
Schema-specific guidelines restrict the freedom to represent criteria as conceptual workinstances to a concrete and commonly agreed ontology that determine what can be saidabout how learning designs should be according to a theoretical standpoint or frame-work These build on an activityndashrole ontology as the one described in the previoussection and add a prescription on the kind of lsquoshould-bersquo predications that are allow-able Table 3 describes examples of a minimal language that could be used for thatpurpose
Semantic learning designs 345
copy 2006 The Author Journal compilation copy 2006 British Educational Communications and Technology Agency
In the examples CAT represents a category of individual expressed through a conceptor a logical formula and the sentence is a synthetic way of expressing some recom-mended or mandatory constraint over such category Other languages could be engi-neered in further research but the one described can be directly expressed in OpenCycand combined with software modules that check or enforce them in some way Notethat this representation does not use direct OpenCyc sentences so that reification is usedinstead of direct logical sentences It is not the aim of this paper to determine the mostpowerful and flexible representation language but only set a foundation for furtherspecification efforts
Analysis and comparison of two theoretical semantic frameworksAs an assessment of the representational capabilities of the scheme just described acase study of representation is sketched in this section We will use the learning designtoolkit described by Conole et al (2004) as the underlying terminology even thoughother frameworks could be used instead
Two rather diverging positions for design will be used as an illustration Table 4 sum-marises the two design scenarios
Table 3 Possible language for use in learning designs
Sentence type Potential uses Example
RecommendationshouldBe (CATSENT)
Recommend kinds ofactivities
Design activities are recommended (GC3)shouldBe(Method subEvent (CATDesignActivity))
ObligationmustBe (CATSENT)
Some activity typesmust follow others
Assessments must follow each expositionto contents (GD1) mustBe(and(ExpositiveActivityIndividualActivity)startAfterEndingOf(CAT
AssessmentActivity)
Activities must havesome characteristic
Every activity must entail interactionmustBe(LDActivity(usedIn(CAT
InteractiveService)))Every activity must provide acontent-oriented learning object forindividual use (similar to GB1)mustBe(LDActivity(and (usedIn(CAT
LearningObject)(learnerRoleNumber(1))))
Some structure shouldbe enforced in methods
Methods should have a sequential actstructuremustBe(Methodnot ((subEvent(CAT Play))))
CAT = category GC3 = Constructivist Guideline 3 (see Table 2) GD1 = Guideline 1
346 British Journal of Educational Technology Vol 37 No 3 2006
copy 2006 The Author Journal compilation copy 2006 British Educational Communications and Technology Agency
The following subsections sketch the main outcomes of an example learning designprocess for both scenarios assuming that the process described by Conole et al (2004)is used in both situations
Differences in modelling the design processThe forces driving the design process in both scenarios are fairly diverging The designof Scenario A will typically be constrained by a clear target specified in terms of anaggregated competency level while in Scenario B there are no concrete competenciesto be acquired and the main objective of the design activity would be twofold On theone hand the activities are targeted at promoting interchange of ideas in an attemptto stimulate casual convergence of interest which is the main value of the culture ofthe special interest group On the other hand the activities have the generic objectiveof strengthening social ties which is an important element of cohesion for the under-taking of joint projects
Table 5 describes the main inputs and possible activities and their objectives at a highlevel
The changes intended to be achieved in each of the scenarios are stated differently witha formal definition in (A) and a shallow cognitive and social representation in the case
Table 4 Learning design scenarios
Scenario DescriptionClassification according to
Conole et al (2004) toolkit
(A) Systemic-Behaviourist A software development company attempts to designlearning units about newtechnologies or softwareproducts for its developerstaff The design is drivenby a cycle of lsquocompetencygap analysisrsquo that modelsorganisational learning in terms of the aggregated competencies of the employeesmdashsimilar to theone described by Sicilia(2005)
Focuses on the individual where a body of informationforms the basis and rawmaterial for learning Learning is mainly considered as nonreflective ie based on skill learning and memorisation
(B) Sociocultural A special interest group ona concrete research topicaims at developing the knowledge of its membersby sharing experiencesand discussing on themesof common interest
Focuses on the social whereexperience is the main source of learning Conscious reflection on experience is transformed into learning
Semantic learning designs 347
copy 2006 The Author Journal compilation copy 2006 British Educational Communications and Technology Agency
of (B) In addition case (B) requires a model of social network that is largely irrelevantin case (A)
Differences in recording activity rationalesThe approach for the recording of design rationales also differs because of the differentcontexts Table 6 summarises the main elements for each of the scenarios
Conclusions and directions for researchCurrent technology and practice related to the description of reusable digital learningresources has resulted in techniques that allow the description of learning activities of
Table 5 Inputs activities and objectives of two learning design scenarios
Scenario Design inputs Activities and their objectives
(A) Systemic-Behaviourist A formal competency recordfor employees and competencyrequirements expressed in terms of amounts of required competencies (assuming a shared measurement scale) Such requirements are actually derived from strategy or from the contingent requirements of upcoming projects
Competency instances are used to measure the levelof Competency-Spec for the required competencies (or their constituents) A formal match considering the agenda of employees is carried out for determining the target learners LearningObjects forindividual use are selected asthe main elements of theactivities and the criteria forselection are again a match ofthe objectives stated inmetadata (as stated incategory 9 of Learning Object MetadatamdashClassification) and thecompetencies required
(B) Sociocultural The expertise of the special interest group members as stated in the concepts addressed by their publicationsprojects or other activities
Informal group activities are the main approach applicablesince there is not a predefinedgroup structure but a cultureof sharing and socialisation The results of the sharing of ideas can be expressed in terms of oc-knowsAboutpredicates and social ties will be modelled through positive valence FOAF-knows(httpxmlnscomfoaf01)predicates with an added numerical value representingstrength
348 British Journal of Educational Technology Vol 37 No 3 2006
copy 2006 The Author Journal compilation copy 2006 British Educational Communications and Technology Agency
a complex kind including their intended participants educational objectives and exe-cution constraints The representation and recording of the assumptions and guidelinesthat were considered in the creation of such resources would extend current techniquesto capture the rationale of the design itself This in turn will eventually result in theavailability of a large base of empirical data available in which theoretical assumptionsor rules are explicitly connected to the resulting designs as these could be linked withthe actual results and trace of the learning processes that used them The resulting bodyof evidence would represent a critical asset in the development of the theory of learningmediated through computers
This paper has described the first attempt to extend existing learning technology modelswith a schema explicitly targeted at describing learning design rationales The OpenCycknowledge base has been used as the semantic reference from which the ontologicalschema has been specified Concretely an activity-based schema based on IMS LD hasbeen described and theoretical assumptions and codified guidelines have been intro-duced The flexibility of the representation schema has been analysed through therepresentation of two stereotyped theoretical positions
Two main research directions should continue the work described here On the onehand further ontological engineering effort is required to come up with a more com-prehensive and rich schema for learning design rationales On the other hand tool
Table 6 Main elements of the two learning design scenarios
Scenario Approach Examples
(A) Systemic-Behaviourist Formal logical sentencesare used to guarantee that every design followsan organisationalobjective
It is required that the objective of every Method is attached to a Competency-Spec instance as objective
Formal assessment is a requirement foraccountability purposes
Guideline (GD1) is attached to everyDesigningLearningResource instance that produces an Act
(B) Sociocultural The interaction in informal settings can beused to derive social models (Sicilia amp Garciacutea2004)
The interactions in discussion canbe used to measure the degree of lsquoproximityrsquo between members Therationale of the activities can bejustified in terms of SocialTies inside SocialNetworks Further the focus of each groupactivity should be clearly stated anda combination of the increase in collective knowledge and the strengthening of social relations can be used for assessment
Semantic learning designs 349
copy 2006 The Author Journal compilation copy 2006 British Educational Communications and Technology Agency
implementation and record gathering efforts would be required to attain the ultimateobjective of the ontological definitions described so far With such effort an unprece-dented research material could be available in some years providing improved oppor-tunities for scientific inquiry on design as an activity regarding pedagogy
Other potential directions for research may be the description of interpretations of thedesign intentions (for which no rationale was explicitly recorded) the mining or detec-tion of patterns as related to rationale description or the detailed use of theoreticalrationales in agents that support the learning process which represent a category ofrelevant learning support software (Holmes in press)
ReferencesAllert H (2004) Coherent social systems for learningmdashan approach for contextualized and
community-centred metadata Journal of Interactive Media in Education 2004 1Baader F Calvanese D McGuinness D Nardi D amp Patel-Schneider P (Eds) (2003) The
description logic handbook Theory implementation and applications Cambridge CambridgeUniversity Press
Bannister D (1981) Personal construct theory and research method In P Reason amp J Rowan(Eds) Human inquiry a sourcebook of new paradigm research (pp 191ndash200) Chichester UKJohn Wiley and Sons
Berners-Lee T Hendler J amp Lassila O (2001) The semantic web Scientific American 284 534ndash43
Brown A amp Voltz B (2005) Elements of effective e-learning design International Review ofResearch in Open and Distance Learning 6 1
Buzza D Bean D Harrigan K amp Carey T (2005) Learning design repositories adaptinglearning design specifications for shared instructional knowledge Canadian Journal of Learningand Technology 30 3
Conole G Dyke M Oliver M amp Seale J (2004) Mapping pedagogy and tools for effectivelearning design Computers amp Education 43 1ndash2 17ndash33
Dodero J M Aedo I amp Diacuteaz P (2002) Participative knowledge production of learning objectsfor e-books The Electronic Library 20 4 296ndash305
Friesen N (2005) Interoperability and learning objects an overview of e-learning standardiza-tion Interdisciplinary Journal of Knowledge and Learning Objects 1 1 23ndash31
Gruber T (1995) Towards principles for the design of ontologies used for knowledge sharingInternational Journal of HumanndashComputer Studies 43 56 907ndash928
Hatchuel A (2002) Towards design theory and expandable rationality the unfinished programof Herbert Simon Journal of Management and Governance 5 3ndash4 260ndash271
Holmes J (in press) Designing agents to support learning by explaining Computers and EducationIEEE Learning Technology Standards Committee (2002) Learning object metadata (LOM) Final
Draft Standard IEEE 1484121ndash2002IMS (2003) IMS Learning Design Information Model (Version 10) Final Specification 20
January 2003Kaptelinin V Kuutti K amp Bannon L (1995) Activity theory basic concepts and applications
In Blumenthalet al (Eds) Humanndashcomputer interaction Lecture Notes in Computer Science 1015189ndash201
Koper R (2004) Use of the semantic web to solve some basic problems in education increaseflexible distributed lifelong learning decrease teacherrsquos workload Journal of Interactive Mediain Education 2004 6
Lenat D (1995) CYC a large-scale investment in knowledge infrastructure Communications ofthe ACM 38 11 33ndash38
350 British Journal of Educational Technology Vol 37 No 3 2006
copy 2006 The Author Journal compilation copy 2006 British Educational Communications and Technology Agency
McGreal R (2004) Learning objects a practical definition International Journal of InstructionalTechnology and Distance Learning 1 9 21ndash32
Packer M amp Goicoechea J (2000) Sociocultural and constructivist theories of learningontology not just epistemology Educational Psychologist 35 4 227ndash241
Palmer J (2001) Fifty major thinkers on education London and New York RoutledgePolsani P R (2003) Use and abuse of reusable learning objects Journal of Digital Information 3
4 Rothwell W amp Kazanas H (1992) Mastering the instructional design process San Francisco CA
Jossey-BassSaacutenchez-Alonso S amp Sicilia M A (2004) Relationships and commitments in learning object
metadata In Proceedings of the 5th International Conference on Information Technology BasedHigher Education and Training ITHET 2004 Istanbul Turkey
Sicilia M A (2005) Ontology-Based Competency Management Infrastructures for theKnowledge-intensive Learning Organization In Lytras amp Naeve (Eds) Intelligent Learning Infra-structures in Knowledge Intensive Organizations A Semantic Web perspective IDEA USA pp 302ndash324
Sicilia M A amp Garciacutea E (2004) Fuzzy group models for adaptation in cooperative informationretrieval contexts Lecture Notes in Computer Science 2932 324ndash334
Sicilia M A amp Garciacutea E (2005) On the convergence of formal ontologies and standardizede-learning Journal of Distance Education Technologies 3 2 13ndash29
Sicilia M A amp Lytras M (2005) On the representation of change according to different ontol-ogies of learning International Journal of Learning and Change 1 1 66ndash79
Sicilia M A Lytras M Rodriacuteguez E amp Garciacutea E (in press) Integrating descriptions of knowl-edge management learning activities into large ontological structures a case study Data andKnowledge Engineering
Tennant M (1991) Expertise as a dimension of adult development New Education 13 1 49ndash55Vanderdonckt J (1999) Development milestones towards a tool for working with guidelines
Interacting with Computers 12 2 81ndash118Wand Y amp Weber R (1995) On the deep structure of information systems Information Systems
Journal 5 203ndash223Wiley D A (2001) The instructional use of learning objects Bloomington IN Association for
Educational Communications and Technology
Semantic learning designs
333
copy 2006 The Author Journal compilation copy 2006 British Educational Communications and Technology Agency
More concretely the modelling of the rationale of learning designs would provide anumber of benefits for researchers and educators in general Among these benefitsSicilia and Lytras (2005) provided the following list (1) the
linking of theoretical assump-tions to practical learning designs
for informative purposes (2) the use of such
links asresources for education of learning designers
(3) the search for
patterns in design situations
that include some given theoretical assumptions and (4) the lsquoinferencersquo of
detailed com-parisons between the effectiveness and adequacy of learning designs
(for similar settings)These aspects are reflections of the practical benefits of the main objective of the record-ing of design rationales which is no other than advancing the scientific inquiry onlearning by enabling the recording of explicit links from design rules to concretedesigns and the latter to the outcomes of the lsquoinstantiationrsquo of these designs in possiblydiverse institutions or groups This would come up with an unprecedented empiricalbase for meta-analysis and knowledge discovery For example hypotheses on the ade-quacy of group-based or individual learning for some given conditions could be testedwith a large empirical base containing the history of interaction and results of multiplesimilar learning programs
This paper approaches the problem of the metadata representation of assumptions andhypothesis related to learning designs in a general sense Because the entities to berepresented are relevant rules or rationales involved in the intellectual process of design-ing learning experiences the schema described is deliberately open-ended This is arequirement imposed by the fact that theories and guidelines related to learning arebased on models that are abstract in nature This in turn is because they attempt tocharacterise learning as a general human cognitive process but there exist at least somerecurring elements that are common to many practical learning designs We have notattempted to build an ontology of existing theories of learning but addressing somelsquoupperrsquo elements that cross-cut any of them eg notions of
change
are one of these upperelements (Sicilia amp Lytras 2005) In any case the fact that diverging accounts oflearning exist is far from controversial and even modern theories somewhat diverge intheir basic ontological commitments (Packer amp Goicoechea 2000)
The representation described departs from a model based on IMS LD but expressed in aricher ontological language Ontology engineering (Gruber 1995) provides a methodto develop semantic conceptualisations and
Semantic Web
technology (Berners-LeeHendler amp Lassila 2001) aims at providing tools and techniques to develop softwarethat exploits them in the context of a learning system Formal ontology provides theknowledge representation infrastructure for ontologies of learning to a level of consid-erable complexity and richness as supported by description logics (Baader CalvaneseMcGuinness Nardi amp Patel-Schneider 2003) It should be noted that the lsquoontology oflearningrsquo explicitly or tacitly assumed actually constraints and drives the behaviourof the system as a whole and the criteria used for the evaluation of the outcomes ofactivities also become a reflection of previous ontological commitments For examplean ontology not considering social issues will not allow (or will not care about) thetracking of the evolution that takes place in the social network of the learners involvedas the activities are carried out
334
British Journal of Educational Technology Vol 37 No 3 2006
copy 2006 The Author Journal compilation copy 2006 British Educational Communications and Technology Agency
The resulting learning design descriptions are called here lsquo
semantic learning design
rsquo torepresent the fact that they are described in ontological terms and that the meaning ofthe design as conceived by the designer is explicitly represented This carries one stepfurther the current metadata descriptions provided by specifications such as LOMSCORM or LD
As a technique for validating the semantic precision of the schema described hereexplicit links have been provided to concepts and relations that are already described ina large upper ontology concretely the
OpenCyc
09 knowledge base This is an alterna-tive to analysis techniques such as the BungendashWandndashWeber (Wand amp Weber 1995)that fosters the reuse of existing open knowledge engineering The subsequent mappingto modern Web-enabled ontology languages such as OWL Web Ontology Language(httpwwww3orgTRowl-ref) is a straightforward step In addition some represen-tative examples of different theoretical standpoints are described as illustrations
The rest of this paper is structured as follows Section 2 provides an ontological accountof an activity-based model of learning activities as supported by IMS LD which consti-tutes the
artefact
that was created by an intellectual process Then Section 3 describesa proposal for an ontological schema targeted at recording design rationales in a broadsense The scheme described is then used in the fourth section to provide as a case studythe comparison of two theoretical frameworks that lead to two different descriptions fora concrete learning design Finally conclusions and future research directions are pro-vided in the fifth section
Recording educational metadata and semantic annotation of learning activities
The first step in the modelling of learning design rationales is that of representing theresults or
artefacts
of the design process In this section the main elements of the IMSLD schema are expressed in ontological terms considering both the design in itself (iethe equivalent to a learning design description) and the entities that result from theapplication of designs in actual occurrences of learning following them (ie the record-ing of actual executions of the planned design in concrete instructional or trainingsettings)
A significant amount of reuse in terminological structures can be achieved by buildingconceptualisations on top of existing large terminological bases like
OpenCyc
(httpwwwopencycorg)
OpenCyc
is the open source version of the
Cyc
knowledge base(Lenat 1995) which contains over 100 000 atomic terms and is provided with anassociated efficient inference engine
Cyc
uses as its underlying definition language avariant of predicate calculus called
CycL
and it attempts to provide a comprehensiveupper ontology of lsquocommon sensersquo knowledge In the rest of the paper the ontologicaldefinitions provided are explicitly linked to
OpenCyc
definitions (in which ontologicaldefinitions of terms and relations are provided in courier font and those that are definedin
OpenCyc
are prefixed by lsquoocrsquo) as a means of reusing existing ontological engineeringefforts and also as a way of validating the concepts being represented
Semantic learning designs
335
copy 2006 The Author Journal compilation copy 2006 British Educational Communications and Technology Agency
Describing the main structural elements of IMS LD
The conceptual model of IMS LD (IMS 2003) provides a concise lsquometalanguagersquo thatcould capture a diversity of pedagogical approaches In doing so it essentially adheresto an activity-centred paradigm (Allert 2004) It should be noted that this model isneutral to specific learning theories Even though it integrates the hierarchical struc-ture of activities and mediation principles of
Activity Theory
(Kaptelinin Kuutti amp Ban-non 1995) it is actually a metalanguage that does not make assumptions about otherelements of such theory as the internalisationexternalisation differentiation
The first step in developing an ontological schema for IMS LD must be that of clarifyingthe main entities to be represented
OpenCyc
provides a rich framework to describetemporal activities concretely the
oc_PurposefulAction
concept (a kind of
oc_Event
) defined as lsquoan action consciously volitionally and purposefully done by atleast one actorrsquo captures the essence of learning activities which are the ultimate targetof the specification Nonetheless IMS LD metadata does not attempt to representing theactivities as actually enacted by persons but they describe
blueprints
that will eventuallybe bound to one or several instantiations of activities and they are intended to besearched as such (Buzza Bean Harrigan amp Carey 2005) In contrast
oc_Event
s andtheir associated entities are related to instantiations and not to their specificationsFurther the concept of
oc_ActorSlot
represents the specific roles played by personsin events so that this could be used to represent the LD
staff
and
learner
role categorybut they do not represent roles understood as templates for their eventual binding toconcrete people
An alternative to representing LD specifications as blueprints is that of creating
Event
types for each activity defined in a learning design But this approach entails at leasttwo problems On the one hand this is not ontologically correct because it does notrepresent the reality that the production of IMS LD is an act of creating specificationartefacts instead of a process of classification of possible real world entities In factmany learning designs will probably be never executed and some of them may bepurposefully created as mere design examples On the other hand the direct creationof learning designs as ontology terms entails that the activities actually performed mustadhere strictly to the original learning design used which severely restricts the flexibil-ity in using the design with ad hoc changes or modifications (or even modified
versions
)This second practical flaw suggests that LD specifications should be represented asseparated entities
This has led us to the major decision of modelling learning designs as specificationsfollowing the definition of
oc_Specification
as lsquoan abstract work that constitutesa description of the properties of a
Situation
or a
SomethingExisting
and sometimes evenentire collections of such thingsrsquo Table 1 summarises the representation of
LearningDesigns
in terms of specifications (specialisations of
oc_Specification
) their mainrelated properties and the implications of each of the specifications when producingactual activities Notifications and completion conditions are omitted for brevityThe concepts of lsquo
learning design
rsquo and lsquo
method
rsquo can be considered as equivalent from an
336
British Journal of Educational Technology Vol 37 No 3 2006
copy 2006 The Author Journal compilation copy 2006 British Educational Communications and Technology Agency
Table 1 Summary of the main specification definitions in the ontological schema
Concept Main descriptions Consequences in actual activities
Method-Spec Objectives and prerequisites(described in the followingsection) Structured in
play
specifications or directly
act
specifications
Instantiation is an independentmeaningful learning activity Playsare cotemporal
oc_subEvent
ssubsumed in the method Note thatthe semantic definition of subeventsyet entails the subsuming of subeventsin the containing event
Play-Spec Collection of sequential acts thatcomprise the play
Because plays are potentiallyconcurrent there are no furthertemporal restrictions
Act-Spec Collection of concurrent role parts that detail the activitiesincluded in the design Role partsand environments are the mainelements related to acts
The sequence of instantiations of actscould be sequenced through the
oc_startsAfterEndingOf
RolendashPart-Spec The linking of role specificationsto concrete activities or activitystructures
This entity is not required in theontological schema because theparticipation of roles in the activitiescan be stated in terms of specificationsof
oc_ActorRole
s or subeventsthat are not of a temporal nature
Activity-Spec The specification of learning andsupport structures in terms of
Role-Spec
s
ibtUsed LearningObject
s and used
Service
s Concrete objectivescan also be attached
Each instantiation of an
Activity-Spec
will result ininstances of specific types of
oc_Activity
Following LD specs
support
activities instantiate for everyrole mapping while
learning
onesinstantiate for all the role mappings
Environment These are descriptions of learning objects and services that do not require anexplicit mapping in the ontological schema
Service Service facilities have to belsquoinstantiated by a local run-timeservicesrsquo eg a chat or discussionforum Their computationalnature can be described interms of concrete
oc_ComputerProgramCW
Subcategories of such conceptualrepresentations of softwareentities can be used for thetypology of services
The
oc_ComputerProcessRunning
term is used to describe executionsof
oc_ComputerProgramCW
which cover every instance of aservice as understoodin IMS LD Thelinking between conceptual andactual running instances is coveredby the required predicate
oc-programOfProcess
Semantic learning designs
337
copy 2006 The Author Journal compilation copy 2006 British Educational Communications and Technology Agency
Learning object Those descriptions provided bylearning object metadataLearning objects can be definedas
oc_IBT
s as described inSicilia Lytras Rodriacuteguez andGarciacutea (in press) and approachesto describing metadata throughontologies yet existmdashsee an overview in Sicilia and Garciacutea(2005)
The learning objects selected arerepresented as instances in theontology which are declared to beused in activities through the
ibtUsed
predicate
Role-Spec Describe stereotypical actors oflearning and support activitiesThese act as lsquocontractrsquo for theactual participants in the senseprovided by Saacutenchez-Alonso andSicilia (2004)
Role specs are mapped to actualusers when the activities take placeThis is mapped to predicates of theclass
oc_ActorRole
Concept Main descriptions Consequences in actual activities
LD
=
Learning Design
Table 1 Continued
ontological perspective as the former allow onemdashand only onemdashinstance of the latterwith no particular logical differences
As described in Table 1 the
OpenCyc
built-in support for modelling temporal events canbe used to express the coordination relationships between the constituents of the unitsof learning The actual instances of learning events will be modelled as instances of
oc_Learning
(lsquothe collection of all events brief or extended in which an agent isacquiring information or know-howrsquo) and more specifically as instances of a speciali-sation
ComputerMediatedLearning
simply requiring the use of information net-works for at least part of the learning activities Because events in
OpenCyc
are alsosituations they have components in addition to the temporal extent including peopleinvolved information things used and other arrangements
Figure 1b depicts the main elements that describe actual current or past enactmentsof activities The schema is based on a clear separation of the specifications and theactual activities that by itself constitutes a metadata layering
OpenCyc
predicatesas
oc_situationConformsToSpecification
allows the linking of situations(every action is a situation also) to the specifications they conformmdashbe it purposely ornot The
LD-Activity term has been introduced as the common category of learningand support activities Methods plays and acts become special subcategories of LD-Activity and the constituent activities of Acts are instances of LD Activity also Rolesare attached to any kind of LD Activity even though it is required for the low-levelstructure The learning design concept of lsquoactivity structurersquo as an aggregation ofactivities does not require an explicit modelling because of the generic mereology ofactivities engineered in OpenCyc
338 British Journal of Educational Technology Vol 37 No 3 2006
copy 2006 The Author Journal compilation copy 2006 British Educational Communications and Technology Agency
Figure 1 (a) Summary of the competency components in the ontological schema (b) Summary of the main activityndashrole definitions in the ontological schema
oc_WorkingEvent
oc_IntelligentAgent
The competency elements aremeasurable characteristics andthey will be typed by the relatedSpecification classes eg Skill
SpecputlnPracticeln
(a)
Competency Competency-Spec
Competencyelement
Skill oc_ Competence
Attitude
Knowledge
has
thingSpecified
oc_Learning
ComputerMediatedLearning
oc_Role
(b)
Staff
Learner
oc_Specification
Method-Spec
Play-Spec
Act-Spec
Support
LD-Activity
contains
contains
Method
Play
Act
startAfterEndingOf
subAction
subAction
Activity-Spec
ltltActorSlotsgtgt
situationConformsToSpec
Activity-Spec
Activity
Semantic learning designs 339
copy 2006 The Author Journal compilation copy 2006 British Educational Communications and Technology Agency
The concept of oc_Role accounts not only for participant roles but also for otherkinds of lsquocomponentsrsquo This is why two specialisations of the concept ofoc_ActorSlot labelled Staff and Learner are required to differentiate partici-pants from other elements As is obvious from the earlier discussion the built-in rep-resentation of activities in OpenCyc is powerful enough for the definition of learningdesign activities
The final part of the ontological specification described so far is that of the mappingfrom LD specifications to the actual past or present activity realisations Theoc_situationConformsToSpecification predicate can be used to trace whichactivity realizations come from a given learning design activity definition and the sameoccurs with the oc_thingSpecified for other learning design elements such asroles or environments Existing learning design execution engines as CopperCore(httpcoppercoreorg) actually map LD specifications to online collaboration so thatthe translation of the ontological representations of learning designs to ontology defi-nitions of activities is straightforward following similar representation
Describing objectives outcomes and preconditionsAccording to IMS learning design learning objectives and prerequisites lsquocan bedescribed using the IMS Reusable Definition of Competency or Educational Objective(RDCEO) format but can also refer to simple resources (eg a text) with a descriptionof the learning objectiversquo As the latter option is not useful from the viewpoint of com-putational semantics we will adhere to the former As commented earlier the compe-tency view is not exclusive to other kinds of objectives related to other ontologicalstandpoints (Sicilia amp Lytras 2005) and other conceptions of change or improvementmay include social ties cognitive structures or the combination of the knowledge of agroup In fact competencies adhere mostly to behaviouristic frameworks that focus onobservable external behaviour (Tennant 1991) rather than on cognitive representa-tions Nonetheless the further discussion is equally applicable to other kinds of objec-tives so that we will adhere to the competency view
oc_Competence in OpenCyc is defined as a quantity that lsquofor describing the level ofskill with which an agent performs some taskrsquo This does not remove the lack of preci-sion that is yet present in RDCEO definitions as described by Sicilia (2005) As analternative Figure 1a provides a simplified account of the model described by Sicilia asrelated to OpenCyc terms in which competence is integrated as one of the elements of thebroader notion of competency according to the model of Rummel (Rothwell amp Kazanas1992) which considers work situations attributes of the individual that are applied insuch situations and measurement with respect to some standard of performance Themost relevant distinction is again the separation of actual competency representationsand competency lsquodefinitionsrsquo (ie specifications) intended to provide a type structure tocompetencies to be used in objective or precondition specification In this case thegeneric oc-thingSpecified predicate can be used for the linking of both levels ofrepresentation
340 British Journal of Educational Technology Vol 37 No 3 2006
copy 2006 The Author Journal compilation copy 2006 British Educational Communications and Technology Agency
Objectives and preconditions can then be expressed as formulas on the acquisition ofcompetencies or their components (competency elements) Descriptions specific toOpenCyc may use oc_CycLSentence-Assertible but simpler enumeration ofcompetency components could be used instead
Describing learning design rationales theories hypotheses guidelines and rulesThe concept of lsquodesignrsquo for our purposes is properly captured in the 5a definitionprovided in the Merriam Webster Online Dictionary (wwwm-wcom) lsquoan underlyingscheme that governs functioning developing or unfoldingrsquo In consequence learningdesigns refer to artefacts and not to the process of devising the educational solutionThis is equivalent to the concept oc_Design subsumed by the notion ofoc_Specification which is described as lsquoan abstract work that constitutes adescription of the properties of a Situation or a SomethingExisting and sometimeseven entire collections of such thingsrsquo In consequence the term LearningDesignwould simply delineate a subset of specifications that are oriented to describe howlearning experiences with concrete objectives should be Trivially oc_Designerdescribes the authors of the learning designs and oc_Designing represent theoc_HumanActivity that leads to the production of learning designs Designing as aspecial kind of activity is carried out (oc_doneBy) by a designer which provides a wayto attach metainformation about the rationale of the design
From an ontological perspective the departing problem in modelling learning designrationales is that their assumptions and hypotheses of different theoretical standpointsneed not be compatible or need not consider the same aspects of reality (Conole et al2004 Packer amp Goicoechea 2000) Furthermore different pedagogical approaches areknown to result in different mappings to activities (Conole et al 2004) This wouldeither require the provision of separate ontologies or the use of a representationalmechanism that allows such kind of potential inconsistency or divergence The conceptof microtheory in OpenCyc provides such representational mechanism intended to orga-nise assertions that depend on lsquoshared set of assumptions on which the truth of theassertions dependsrsquo Definitions inside the same microtheory need to be consistent butthis is not required across microtheories The descriptions provided in the earlier sectioncan be used as a base microtheory called LearningActivitiesMt for others that are spe-cific to concrete theoretical frameworks In this section we will describe elements thatare common to the description of theoretical frameworks the distinction among micro-theories will not be discussed
The intellectual processes of learning design in themselves will be represented asoc_Designing activities carried out by oc_Designers An additional consider-ation is that the information about the rationale of the design would be better attachedto representations of each of the design processes (ie to oc_Designing instances)than to the final design This enables the registering of different assumptions for differ-ent contributions to the same artefact eg different revisions from the same of differentdesigners
Semantic learning designs 341
copy 2006 The Author Journal compilation copy 2006 British Educational Communications and Technology Agency
Several designers may participate in the same designing activity and formal frameworksfor collaborative learning resource creation (Dodero Aedo amp Diacuteaz 2002) can be inte-grated by clearly specifying which class of things can be the outcome of a designingprocess For the sake of flexibility the concept of LearningDesignPart is introduced as anall-encompassing category that includes each of the elements described in the previoussection This includes not only activities and roles but also objectives conditions andthe like Furthermore the granularity of the descriptions allows for different levels fromconsidering a unique designing activity for the whole package to recording each ele-ment of the learning design comprehensively Because oc_Actions are subsumed bythe concept of oc_Event considered as a lsquodynamic situation in which the state of theworld changesrsquo each edition of a metadata record or any other form of description maybe subject to be described The concept is flexible enough to be able to record fine-grained detailed decisions This of course may include different versions of the sameartefacts but this is an orthogonal aspect of configuration that is not relevant to ourpresent discussion
Figure 2 depicts the main elements of the loci to which rationales assumptions andguidelines will be attached
A LearningDesigner can be defined as the class of humans that performactivities of learning resource design The consideration of learning design as aunique human activity could be considered as restrictive of intelligent softwarefor the task Simply changing the oc_Designer class to a subconcept ofoc_IntelligentAgent could remove this but this entails a change in Open-Cyc knowledge base The concept oc_Designing is a subclass of the conceptoc_ThinkingAboutTheStateOfTheWorld which represents lsquothinking pro-cessesrsquo which include mental events such as planning evaluating or inventingThus the concept of design described so far should be understood in terms of the
Figure 2 Main elements of the ontological schema for the loci of design rationales
oc_performedBy modifies
342 British Journal of Educational Technology Vol 37 No 3 2006
copy 2006 The Author Journal compilation copy 2006 British Educational Communications and Technology Agency
concept of lsquoexpanded rationalityrsquo as described by Hatchuel (2002) integrating cre-ativity and unexpected expansions of the original requirements This precludes onto-logical definitions in which the problem space is completely bounded a priori Inconsequence a degree of openness is necessary to integrate different kinds of detailin description from fully described ones to others with shallower semantics eg someproviding only references to generic assumptions
Theoretical frameworksWe will make the distinction that guidelines (as further defined) are the propositions thatdirectly predicate about the validity or appropriateness of learning design parts whileother theoretical propositions about learning that are not directly saying anythingabout the design of digital learning resources are considered as assumptions followingthe general sense provided in the Merriam-Webster dictionary lsquoa fact or statement (asa proposition axiom postulate or notion) taken for grantedrsquo
The fact that some assumption hypotheses or proposed law about learning is sustainedby a broader empirical base or is better sustained by rational means is out of the scopeof this work because here we focus on design practice (even though it is of an obviousinterest for researchers) In principle even unorthodox positions may be sustainedbecause the mechanism of microtheories and their inclusion mechanism allows forgreat flexibility in modelling them
Table 2 provides some examples of theoretical assumptions and related guidelines forlsquoextremersquo archetypical standpoints on learning These are not intended to reflect actualcommon positions but they are provided as illustrative examples of radical positionsthat could be modelled with our ontological schema
The examples provided in Table 2 go a step further than toolkits in Conole et al (2004)in making the design rationale explicit Nonetheless in many cases lsquocommon sensersquo isapplied to design and a shallower approach to specification could be used instead egproviding simply links to the theoretical standpoints used It should be noted also thatthe resulting guidelines vary greatly in their level of concreteness and some of themmay be considered as mandatory from some standpoints while others are mere generalrepresentations This would lead to different computational strategies
Guidelines and rulesThe WordNet (httpwordnetprincetonedu) thesaurus defines guidelines as lsquoguidancerelative to setting standards or determining a course of actionrsquo or lsquoa rule or principlethat provides guidance to appropriate behaviorrsquo In our case guidelines are statementsthat serve as criteria for decision in designing learning Given that e-learning designis still far from being a mature discipline guidelinesmdashas instances of recordedknowledgemdashcan be characterised as oc_NormativeSpecifications becausethey describe how learning resource design activities and their outcomes lsquoshould bersquoin a propositional form Propositions lsquohave some truth value in some context or
Semantic learning designs 343
copy 2006 The Author Journal compilation copy 2006 British Educational Communications and Technology Agency
ldquoworldrdquo rsquo Furthermore propositions lsquoare assumed to be representable (at least in prin-ciple) by a sentence in some formal or natural languagersquo This raises the need for alanguage that expresses guidelines in a general sense There are two main options forsuch representation be them formal logical sentences or reified guidelines
Formal logical sentence representation entails the formalisation through sentences ofthe constraints of the designs For example if lsquoevery learning resource must be activersquoaccording to some theoretical standpoint (eg one that fosters reflective learning) thissentence may be represented as an axiom and put into the appropriate separate micro-theory as the following CycL fragment
($forAll X ($implies
($isa X $LearningObject)
($resourceType X $Active))
Then resources that are not active are simply excluded from the notion itself This is astrong position in the sense that precludes some views on the learning design process
Table 2 Some examples of theoretical assumptions and guidelines
Archetypicalstandpoint Assumptions Guidelines
Instructivist(Behaviouristic)
bull Learning occurs through association and reinforcement
bull The results of learning can be appreciated through observable outcomes
(GB1) Every instance of oc-Learning entails theexposure to some LearningObjectthat produces the association (GB2) Learning Activitysequences should provide assessmentactivities that confirm theassociations that should have beenestablished so far
Constructivist bull Learning proceeds by building mental structures
(GC1) Agentrsquos mental structuresshould be represented(GC2) These structures should bepre-assessed in order to becomparable with the resulting ones(GC3) Design Activities arerecommended(GC4) Contrast Activities arerecommended
Sociocultural bull Learning occurs through social participatio
(GSC1) Every instance ofoc-Learning entails theinteraction of severaloc-IntelligentAgents asoc-Actors
344 British Journal of Educational Technology Vol 37 No 3 2006
copy 2006 The Author Journal compilation copy 2006 British Educational Communications and Technology Agency
Another typical example may be that of a sociocultural guideline that preclude Methodsthat do not entail multirole activities that provide access to social interaction ServicesThis approach to guidelines provides the best framework for automation and wouldenable the creation of microtheories that represent clearly formulated views onlearning Nonetheless there are two major problems for the use of this option as theunique mechanism for encoding design rationales
1 There is no clear-cut distinction between models and theories of learning becausethey come from a history of schools of thought with interrelations (Palmer 2001)This would lead to the need for the representation of dozens of variants reflectingpositions that only diverge in some aspects
2 Many guidelines cannot be represented as logical sentences because theyrepresent lsquorecommendationsrsquo or vague requirements eg see (GC3) or (GC4)(Table 2)
These issues suggest that a combination of formal sentences with reified guidelinesprovide the best option for flexibility The representation of guidelines has beenstudied in the field of HumanndashComputer-Interaction (HCI) in the last decade(Vanderdonckt 1999) and the main philosophy of such representations remainsvalid in the learning design domain The major challenge for such representations isthat of providing computational semantics that could be used for the implementationof software support but this is difficult to attain in all the cases Thus the concept oflearning design guideline will provide two kinds of representations general andschema-specific
The general use of instances of oc_NormativeSpecification allows the model-ling of lsquoconceptual works that describe how something should bersquo This may includerepresentation of known normative representations or procedures For example onemay refer to a concrete learning design model as that derived from Bannister (1981)which states that creativity cycles follow an initial phase of brainstorming (idea cap-ture) followed by a phase in which key issues are focused into an operation strategy(idea development) and a third phase in which project control takes place through alsquorecursive cyclersquo (operational management) The reference could be made by pointing tothe academic reference describing it as a conceptual work or it could be described asan Activity-Spec modelled explicitly inside the ontology As another examplesome detailed guidelines could be modelled by the normative specification sub-category of oc-Instructions defined as lsquoa sequence of tasks to be performed byan agentrsquo
Schema-specific guidelines restrict the freedom to represent criteria as conceptual workinstances to a concrete and commonly agreed ontology that determine what can be saidabout how learning designs should be according to a theoretical standpoint or frame-work These build on an activityndashrole ontology as the one described in the previoussection and add a prescription on the kind of lsquoshould-bersquo predications that are allow-able Table 3 describes examples of a minimal language that could be used for thatpurpose
Semantic learning designs 345
copy 2006 The Author Journal compilation copy 2006 British Educational Communications and Technology Agency
In the examples CAT represents a category of individual expressed through a conceptor a logical formula and the sentence is a synthetic way of expressing some recom-mended or mandatory constraint over such category Other languages could be engi-neered in further research but the one described can be directly expressed in OpenCycand combined with software modules that check or enforce them in some way Notethat this representation does not use direct OpenCyc sentences so that reification is usedinstead of direct logical sentences It is not the aim of this paper to determine the mostpowerful and flexible representation language but only set a foundation for furtherspecification efforts
Analysis and comparison of two theoretical semantic frameworksAs an assessment of the representational capabilities of the scheme just described acase study of representation is sketched in this section We will use the learning designtoolkit described by Conole et al (2004) as the underlying terminology even thoughother frameworks could be used instead
Two rather diverging positions for design will be used as an illustration Table 4 sum-marises the two design scenarios
Table 3 Possible language for use in learning designs
Sentence type Potential uses Example
RecommendationshouldBe (CATSENT)
Recommend kinds ofactivities
Design activities are recommended (GC3)shouldBe(Method subEvent (CATDesignActivity))
ObligationmustBe (CATSENT)
Some activity typesmust follow others
Assessments must follow each expositionto contents (GD1) mustBe(and(ExpositiveActivityIndividualActivity)startAfterEndingOf(CAT
AssessmentActivity)
Activities must havesome characteristic
Every activity must entail interactionmustBe(LDActivity(usedIn(CAT
InteractiveService)))Every activity must provide acontent-oriented learning object forindividual use (similar to GB1)mustBe(LDActivity(and (usedIn(CAT
LearningObject)(learnerRoleNumber(1))))
Some structure shouldbe enforced in methods
Methods should have a sequential actstructuremustBe(Methodnot ((subEvent(CAT Play))))
CAT = category GC3 = Constructivist Guideline 3 (see Table 2) GD1 = Guideline 1
346 British Journal of Educational Technology Vol 37 No 3 2006
copy 2006 The Author Journal compilation copy 2006 British Educational Communications and Technology Agency
The following subsections sketch the main outcomes of an example learning designprocess for both scenarios assuming that the process described by Conole et al (2004)is used in both situations
Differences in modelling the design processThe forces driving the design process in both scenarios are fairly diverging The designof Scenario A will typically be constrained by a clear target specified in terms of anaggregated competency level while in Scenario B there are no concrete competenciesto be acquired and the main objective of the design activity would be twofold On theone hand the activities are targeted at promoting interchange of ideas in an attemptto stimulate casual convergence of interest which is the main value of the culture ofthe special interest group On the other hand the activities have the generic objectiveof strengthening social ties which is an important element of cohesion for the under-taking of joint projects
Table 5 describes the main inputs and possible activities and their objectives at a highlevel
The changes intended to be achieved in each of the scenarios are stated differently witha formal definition in (A) and a shallow cognitive and social representation in the case
Table 4 Learning design scenarios
Scenario DescriptionClassification according to
Conole et al (2004) toolkit
(A) Systemic-Behaviourist A software development company attempts to designlearning units about newtechnologies or softwareproducts for its developerstaff The design is drivenby a cycle of lsquocompetencygap analysisrsquo that modelsorganisational learning in terms of the aggregated competencies of the employeesmdashsimilar to theone described by Sicilia(2005)
Focuses on the individual where a body of informationforms the basis and rawmaterial for learning Learning is mainly considered as nonreflective ie based on skill learning and memorisation
(B) Sociocultural A special interest group ona concrete research topicaims at developing the knowledge of its membersby sharing experiencesand discussing on themesof common interest
Focuses on the social whereexperience is the main source of learning Conscious reflection on experience is transformed into learning
Semantic learning designs 347
copy 2006 The Author Journal compilation copy 2006 British Educational Communications and Technology Agency
of (B) In addition case (B) requires a model of social network that is largely irrelevantin case (A)
Differences in recording activity rationalesThe approach for the recording of design rationales also differs because of the differentcontexts Table 6 summarises the main elements for each of the scenarios
Conclusions and directions for researchCurrent technology and practice related to the description of reusable digital learningresources has resulted in techniques that allow the description of learning activities of
Table 5 Inputs activities and objectives of two learning design scenarios
Scenario Design inputs Activities and their objectives
(A) Systemic-Behaviourist A formal competency recordfor employees and competencyrequirements expressed in terms of amounts of required competencies (assuming a shared measurement scale) Such requirements are actually derived from strategy or from the contingent requirements of upcoming projects
Competency instances are used to measure the levelof Competency-Spec for the required competencies (or their constituents) A formal match considering the agenda of employees is carried out for determining the target learners LearningObjects forindividual use are selected asthe main elements of theactivities and the criteria forselection are again a match ofthe objectives stated inmetadata (as stated incategory 9 of Learning Object MetadatamdashClassification) and thecompetencies required
(B) Sociocultural The expertise of the special interest group members as stated in the concepts addressed by their publicationsprojects or other activities
Informal group activities are the main approach applicablesince there is not a predefinedgroup structure but a cultureof sharing and socialisation The results of the sharing of ideas can be expressed in terms of oc-knowsAboutpredicates and social ties will be modelled through positive valence FOAF-knows(httpxmlnscomfoaf01)predicates with an added numerical value representingstrength
348 British Journal of Educational Technology Vol 37 No 3 2006
copy 2006 The Author Journal compilation copy 2006 British Educational Communications and Technology Agency
a complex kind including their intended participants educational objectives and exe-cution constraints The representation and recording of the assumptions and guidelinesthat were considered in the creation of such resources would extend current techniquesto capture the rationale of the design itself This in turn will eventually result in theavailability of a large base of empirical data available in which theoretical assumptionsor rules are explicitly connected to the resulting designs as these could be linked withthe actual results and trace of the learning processes that used them The resulting bodyof evidence would represent a critical asset in the development of the theory of learningmediated through computers
This paper has described the first attempt to extend existing learning technology modelswith a schema explicitly targeted at describing learning design rationales The OpenCycknowledge base has been used as the semantic reference from which the ontologicalschema has been specified Concretely an activity-based schema based on IMS LD hasbeen described and theoretical assumptions and codified guidelines have been intro-duced The flexibility of the representation schema has been analysed through therepresentation of two stereotyped theoretical positions
Two main research directions should continue the work described here On the onehand further ontological engineering effort is required to come up with a more com-prehensive and rich schema for learning design rationales On the other hand tool
Table 6 Main elements of the two learning design scenarios
Scenario Approach Examples
(A) Systemic-Behaviourist Formal logical sentencesare used to guarantee that every design followsan organisationalobjective
It is required that the objective of every Method is attached to a Competency-Spec instance as objective
Formal assessment is a requirement foraccountability purposes
Guideline (GD1) is attached to everyDesigningLearningResource instance that produces an Act
(B) Sociocultural The interaction in informal settings can beused to derive social models (Sicilia amp Garciacutea2004)
The interactions in discussion canbe used to measure the degree of lsquoproximityrsquo between members Therationale of the activities can bejustified in terms of SocialTies inside SocialNetworks Further the focus of each groupactivity should be clearly stated anda combination of the increase in collective knowledge and the strengthening of social relations can be used for assessment
Semantic learning designs 349
copy 2006 The Author Journal compilation copy 2006 British Educational Communications and Technology Agency
implementation and record gathering efforts would be required to attain the ultimateobjective of the ontological definitions described so far With such effort an unprece-dented research material could be available in some years providing improved oppor-tunities for scientific inquiry on design as an activity regarding pedagogy
Other potential directions for research may be the description of interpretations of thedesign intentions (for which no rationale was explicitly recorded) the mining or detec-tion of patterns as related to rationale description or the detailed use of theoreticalrationales in agents that support the learning process which represent a category ofrelevant learning support software (Holmes in press)
ReferencesAllert H (2004) Coherent social systems for learningmdashan approach for contextualized and
community-centred metadata Journal of Interactive Media in Education 2004 1Baader F Calvanese D McGuinness D Nardi D amp Patel-Schneider P (Eds) (2003) The
description logic handbook Theory implementation and applications Cambridge CambridgeUniversity Press
Bannister D (1981) Personal construct theory and research method In P Reason amp J Rowan(Eds) Human inquiry a sourcebook of new paradigm research (pp 191ndash200) Chichester UKJohn Wiley and Sons
Berners-Lee T Hendler J amp Lassila O (2001) The semantic web Scientific American 284 534ndash43
Brown A amp Voltz B (2005) Elements of effective e-learning design International Review ofResearch in Open and Distance Learning 6 1
Buzza D Bean D Harrigan K amp Carey T (2005) Learning design repositories adaptinglearning design specifications for shared instructional knowledge Canadian Journal of Learningand Technology 30 3
Conole G Dyke M Oliver M amp Seale J (2004) Mapping pedagogy and tools for effectivelearning design Computers amp Education 43 1ndash2 17ndash33
Dodero J M Aedo I amp Diacuteaz P (2002) Participative knowledge production of learning objectsfor e-books The Electronic Library 20 4 296ndash305
Friesen N (2005) Interoperability and learning objects an overview of e-learning standardiza-tion Interdisciplinary Journal of Knowledge and Learning Objects 1 1 23ndash31
Gruber T (1995) Towards principles for the design of ontologies used for knowledge sharingInternational Journal of HumanndashComputer Studies 43 56 907ndash928
Hatchuel A (2002) Towards design theory and expandable rationality the unfinished programof Herbert Simon Journal of Management and Governance 5 3ndash4 260ndash271
Holmes J (in press) Designing agents to support learning by explaining Computers and EducationIEEE Learning Technology Standards Committee (2002) Learning object metadata (LOM) Final
Draft Standard IEEE 1484121ndash2002IMS (2003) IMS Learning Design Information Model (Version 10) Final Specification 20
January 2003Kaptelinin V Kuutti K amp Bannon L (1995) Activity theory basic concepts and applications
In Blumenthalet al (Eds) Humanndashcomputer interaction Lecture Notes in Computer Science 1015189ndash201
Koper R (2004) Use of the semantic web to solve some basic problems in education increaseflexible distributed lifelong learning decrease teacherrsquos workload Journal of Interactive Mediain Education 2004 6
Lenat D (1995) CYC a large-scale investment in knowledge infrastructure Communications ofthe ACM 38 11 33ndash38
350 British Journal of Educational Technology Vol 37 No 3 2006
copy 2006 The Author Journal compilation copy 2006 British Educational Communications and Technology Agency
McGreal R (2004) Learning objects a practical definition International Journal of InstructionalTechnology and Distance Learning 1 9 21ndash32
Packer M amp Goicoechea J (2000) Sociocultural and constructivist theories of learningontology not just epistemology Educational Psychologist 35 4 227ndash241
Palmer J (2001) Fifty major thinkers on education London and New York RoutledgePolsani P R (2003) Use and abuse of reusable learning objects Journal of Digital Information 3
4 Rothwell W amp Kazanas H (1992) Mastering the instructional design process San Francisco CA
Jossey-BassSaacutenchez-Alonso S amp Sicilia M A (2004) Relationships and commitments in learning object
metadata In Proceedings of the 5th International Conference on Information Technology BasedHigher Education and Training ITHET 2004 Istanbul Turkey
Sicilia M A (2005) Ontology-Based Competency Management Infrastructures for theKnowledge-intensive Learning Organization In Lytras amp Naeve (Eds) Intelligent Learning Infra-structures in Knowledge Intensive Organizations A Semantic Web perspective IDEA USA pp 302ndash324
Sicilia M A amp Garciacutea E (2004) Fuzzy group models for adaptation in cooperative informationretrieval contexts Lecture Notes in Computer Science 2932 324ndash334
Sicilia M A amp Garciacutea E (2005) On the convergence of formal ontologies and standardizede-learning Journal of Distance Education Technologies 3 2 13ndash29
Sicilia M A amp Lytras M (2005) On the representation of change according to different ontol-ogies of learning International Journal of Learning and Change 1 1 66ndash79
Sicilia M A Lytras M Rodriacuteguez E amp Garciacutea E (in press) Integrating descriptions of knowl-edge management learning activities into large ontological structures a case study Data andKnowledge Engineering
Tennant M (1991) Expertise as a dimension of adult development New Education 13 1 49ndash55Vanderdonckt J (1999) Development milestones towards a tool for working with guidelines
Interacting with Computers 12 2 81ndash118Wand Y amp Weber R (1995) On the deep structure of information systems Information Systems
Journal 5 203ndash223Wiley D A (2001) The instructional use of learning objects Bloomington IN Association for
Educational Communications and Technology
334
British Journal of Educational Technology Vol 37 No 3 2006
copy 2006 The Author Journal compilation copy 2006 British Educational Communications and Technology Agency
The resulting learning design descriptions are called here lsquo
semantic learning design
rsquo torepresent the fact that they are described in ontological terms and that the meaning ofthe design as conceived by the designer is explicitly represented This carries one stepfurther the current metadata descriptions provided by specifications such as LOMSCORM or LD
As a technique for validating the semantic precision of the schema described hereexplicit links have been provided to concepts and relations that are already described ina large upper ontology concretely the
OpenCyc
09 knowledge base This is an alterna-tive to analysis techniques such as the BungendashWandndashWeber (Wand amp Weber 1995)that fosters the reuse of existing open knowledge engineering The subsequent mappingto modern Web-enabled ontology languages such as OWL Web Ontology Language(httpwwww3orgTRowl-ref) is a straightforward step In addition some represen-tative examples of different theoretical standpoints are described as illustrations
The rest of this paper is structured as follows Section 2 provides an ontological accountof an activity-based model of learning activities as supported by IMS LD which consti-tutes the
artefact
that was created by an intellectual process Then Section 3 describesa proposal for an ontological schema targeted at recording design rationales in a broadsense The scheme described is then used in the fourth section to provide as a case studythe comparison of two theoretical frameworks that lead to two different descriptions fora concrete learning design Finally conclusions and future research directions are pro-vided in the fifth section
Recording educational metadata and semantic annotation of learning activities
The first step in the modelling of learning design rationales is that of representing theresults or
artefacts
of the design process In this section the main elements of the IMSLD schema are expressed in ontological terms considering both the design in itself (iethe equivalent to a learning design description) and the entities that result from theapplication of designs in actual occurrences of learning following them (ie the record-ing of actual executions of the planned design in concrete instructional or trainingsettings)
A significant amount of reuse in terminological structures can be achieved by buildingconceptualisations on top of existing large terminological bases like
OpenCyc
(httpwwwopencycorg)
OpenCyc
is the open source version of the
Cyc
knowledge base(Lenat 1995) which contains over 100 000 atomic terms and is provided with anassociated efficient inference engine
Cyc
uses as its underlying definition language avariant of predicate calculus called
CycL
and it attempts to provide a comprehensiveupper ontology of lsquocommon sensersquo knowledge In the rest of the paper the ontologicaldefinitions provided are explicitly linked to
OpenCyc
definitions (in which ontologicaldefinitions of terms and relations are provided in courier font and those that are definedin
OpenCyc
are prefixed by lsquoocrsquo) as a means of reusing existing ontological engineeringefforts and also as a way of validating the concepts being represented
Semantic learning designs
335
copy 2006 The Author Journal compilation copy 2006 British Educational Communications and Technology Agency
Describing the main structural elements of IMS LD
The conceptual model of IMS LD (IMS 2003) provides a concise lsquometalanguagersquo thatcould capture a diversity of pedagogical approaches In doing so it essentially adheresto an activity-centred paradigm (Allert 2004) It should be noted that this model isneutral to specific learning theories Even though it integrates the hierarchical struc-ture of activities and mediation principles of
Activity Theory
(Kaptelinin Kuutti amp Ban-non 1995) it is actually a metalanguage that does not make assumptions about otherelements of such theory as the internalisationexternalisation differentiation
The first step in developing an ontological schema for IMS LD must be that of clarifyingthe main entities to be represented
OpenCyc
provides a rich framework to describetemporal activities concretely the
oc_PurposefulAction
concept (a kind of
oc_Event
) defined as lsquoan action consciously volitionally and purposefully done by atleast one actorrsquo captures the essence of learning activities which are the ultimate targetof the specification Nonetheless IMS LD metadata does not attempt to representing theactivities as actually enacted by persons but they describe
blueprints
that will eventuallybe bound to one or several instantiations of activities and they are intended to besearched as such (Buzza Bean Harrigan amp Carey 2005) In contrast
oc_Event
s andtheir associated entities are related to instantiations and not to their specificationsFurther the concept of
oc_ActorSlot
represents the specific roles played by personsin events so that this could be used to represent the LD
staff
and
learner
role categorybut they do not represent roles understood as templates for their eventual binding toconcrete people
An alternative to representing LD specifications as blueprints is that of creating
Event
types for each activity defined in a learning design But this approach entails at leasttwo problems On the one hand this is not ontologically correct because it does notrepresent the reality that the production of IMS LD is an act of creating specificationartefacts instead of a process of classification of possible real world entities In factmany learning designs will probably be never executed and some of them may bepurposefully created as mere design examples On the other hand the direct creationof learning designs as ontology terms entails that the activities actually performed mustadhere strictly to the original learning design used which severely restricts the flexibil-ity in using the design with ad hoc changes or modifications (or even modified
versions
)This second practical flaw suggests that LD specifications should be represented asseparated entities
This has led us to the major decision of modelling learning designs as specificationsfollowing the definition of
oc_Specification
as lsquoan abstract work that constitutesa description of the properties of a
Situation
or a
SomethingExisting
and sometimes evenentire collections of such thingsrsquo Table 1 summarises the representation of
LearningDesigns
in terms of specifications (specialisations of
oc_Specification
) their mainrelated properties and the implications of each of the specifications when producingactual activities Notifications and completion conditions are omitted for brevityThe concepts of lsquo
learning design
rsquo and lsquo
method
rsquo can be considered as equivalent from an
336
British Journal of Educational Technology Vol 37 No 3 2006
copy 2006 The Author Journal compilation copy 2006 British Educational Communications and Technology Agency
Table 1 Summary of the main specification definitions in the ontological schema
Concept Main descriptions Consequences in actual activities
Method-Spec Objectives and prerequisites(described in the followingsection) Structured in
play
specifications or directly
act
specifications
Instantiation is an independentmeaningful learning activity Playsare cotemporal
oc_subEvent
ssubsumed in the method Note thatthe semantic definition of subeventsyet entails the subsuming of subeventsin the containing event
Play-Spec Collection of sequential acts thatcomprise the play
Because plays are potentiallyconcurrent there are no furthertemporal restrictions
Act-Spec Collection of concurrent role parts that detail the activitiesincluded in the design Role partsand environments are the mainelements related to acts
The sequence of instantiations of actscould be sequenced through the
oc_startsAfterEndingOf
RolendashPart-Spec The linking of role specificationsto concrete activities or activitystructures
This entity is not required in theontological schema because theparticipation of roles in the activitiescan be stated in terms of specificationsof
oc_ActorRole
s or subeventsthat are not of a temporal nature
Activity-Spec The specification of learning andsupport structures in terms of
Role-Spec
s
ibtUsed LearningObject
s and used
Service
s Concrete objectivescan also be attached
Each instantiation of an
Activity-Spec
will result ininstances of specific types of
oc_Activity
Following LD specs
support
activities instantiate for everyrole mapping while
learning
onesinstantiate for all the role mappings
Environment These are descriptions of learning objects and services that do not require anexplicit mapping in the ontological schema
Service Service facilities have to belsquoinstantiated by a local run-timeservicesrsquo eg a chat or discussionforum Their computationalnature can be described interms of concrete
oc_ComputerProgramCW
Subcategories of such conceptualrepresentations of softwareentities can be used for thetypology of services
The
oc_ComputerProcessRunning
term is used to describe executionsof
oc_ComputerProgramCW
which cover every instance of aservice as understoodin IMS LD Thelinking between conceptual andactual running instances is coveredby the required predicate
oc-programOfProcess
Semantic learning designs
337
copy 2006 The Author Journal compilation copy 2006 British Educational Communications and Technology Agency
Learning object Those descriptions provided bylearning object metadataLearning objects can be definedas
oc_IBT
s as described inSicilia Lytras Rodriacuteguez andGarciacutea (in press) and approachesto describing metadata throughontologies yet existmdashsee an overview in Sicilia and Garciacutea(2005)
The learning objects selected arerepresented as instances in theontology which are declared to beused in activities through the
ibtUsed
predicate
Role-Spec Describe stereotypical actors oflearning and support activitiesThese act as lsquocontractrsquo for theactual participants in the senseprovided by Saacutenchez-Alonso andSicilia (2004)
Role specs are mapped to actualusers when the activities take placeThis is mapped to predicates of theclass
oc_ActorRole
Concept Main descriptions Consequences in actual activities
LD
=
Learning Design
Table 1 Continued
ontological perspective as the former allow onemdashand only onemdashinstance of the latterwith no particular logical differences
As described in Table 1 the
OpenCyc
built-in support for modelling temporal events canbe used to express the coordination relationships between the constituents of the unitsof learning The actual instances of learning events will be modelled as instances of
oc_Learning
(lsquothe collection of all events brief or extended in which an agent isacquiring information or know-howrsquo) and more specifically as instances of a speciali-sation
ComputerMediatedLearning
simply requiring the use of information net-works for at least part of the learning activities Because events in
OpenCyc
are alsosituations they have components in addition to the temporal extent including peopleinvolved information things used and other arrangements
Figure 1b depicts the main elements that describe actual current or past enactmentsof activities The schema is based on a clear separation of the specifications and theactual activities that by itself constitutes a metadata layering
OpenCyc
predicatesas
oc_situationConformsToSpecification
allows the linking of situations(every action is a situation also) to the specifications they conformmdashbe it purposely ornot The
LD-Activity term has been introduced as the common category of learningand support activities Methods plays and acts become special subcategories of LD-Activity and the constituent activities of Acts are instances of LD Activity also Rolesare attached to any kind of LD Activity even though it is required for the low-levelstructure The learning design concept of lsquoactivity structurersquo as an aggregation ofactivities does not require an explicit modelling because of the generic mereology ofactivities engineered in OpenCyc
338 British Journal of Educational Technology Vol 37 No 3 2006
copy 2006 The Author Journal compilation copy 2006 British Educational Communications and Technology Agency
Figure 1 (a) Summary of the competency components in the ontological schema (b) Summary of the main activityndashrole definitions in the ontological schema
oc_WorkingEvent
oc_IntelligentAgent
The competency elements aremeasurable characteristics andthey will be typed by the relatedSpecification classes eg Skill
SpecputlnPracticeln
(a)
Competency Competency-Spec
Competencyelement
Skill oc_ Competence
Attitude
Knowledge
has
thingSpecified
oc_Learning
ComputerMediatedLearning
oc_Role
(b)
Staff
Learner
oc_Specification
Method-Spec
Play-Spec
Act-Spec
Support
LD-Activity
contains
contains
Method
Play
Act
startAfterEndingOf
subAction
subAction
Activity-Spec
ltltActorSlotsgtgt
situationConformsToSpec
Activity-Spec
Activity
Semantic learning designs 339
copy 2006 The Author Journal compilation copy 2006 British Educational Communications and Technology Agency
The concept of oc_Role accounts not only for participant roles but also for otherkinds of lsquocomponentsrsquo This is why two specialisations of the concept ofoc_ActorSlot labelled Staff and Learner are required to differentiate partici-pants from other elements As is obvious from the earlier discussion the built-in rep-resentation of activities in OpenCyc is powerful enough for the definition of learningdesign activities
The final part of the ontological specification described so far is that of the mappingfrom LD specifications to the actual past or present activity realisations Theoc_situationConformsToSpecification predicate can be used to trace whichactivity realizations come from a given learning design activity definition and the sameoccurs with the oc_thingSpecified for other learning design elements such asroles or environments Existing learning design execution engines as CopperCore(httpcoppercoreorg) actually map LD specifications to online collaboration so thatthe translation of the ontological representations of learning designs to ontology defi-nitions of activities is straightforward following similar representation
Describing objectives outcomes and preconditionsAccording to IMS learning design learning objectives and prerequisites lsquocan bedescribed using the IMS Reusable Definition of Competency or Educational Objective(RDCEO) format but can also refer to simple resources (eg a text) with a descriptionof the learning objectiversquo As the latter option is not useful from the viewpoint of com-putational semantics we will adhere to the former As commented earlier the compe-tency view is not exclusive to other kinds of objectives related to other ontologicalstandpoints (Sicilia amp Lytras 2005) and other conceptions of change or improvementmay include social ties cognitive structures or the combination of the knowledge of agroup In fact competencies adhere mostly to behaviouristic frameworks that focus onobservable external behaviour (Tennant 1991) rather than on cognitive representa-tions Nonetheless the further discussion is equally applicable to other kinds of objec-tives so that we will adhere to the competency view
oc_Competence in OpenCyc is defined as a quantity that lsquofor describing the level ofskill with which an agent performs some taskrsquo This does not remove the lack of preci-sion that is yet present in RDCEO definitions as described by Sicilia (2005) As analternative Figure 1a provides a simplified account of the model described by Sicilia asrelated to OpenCyc terms in which competence is integrated as one of the elements of thebroader notion of competency according to the model of Rummel (Rothwell amp Kazanas1992) which considers work situations attributes of the individual that are applied insuch situations and measurement with respect to some standard of performance Themost relevant distinction is again the separation of actual competency representationsand competency lsquodefinitionsrsquo (ie specifications) intended to provide a type structure tocompetencies to be used in objective or precondition specification In this case thegeneric oc-thingSpecified predicate can be used for the linking of both levels ofrepresentation
340 British Journal of Educational Technology Vol 37 No 3 2006
copy 2006 The Author Journal compilation copy 2006 British Educational Communications and Technology Agency
Objectives and preconditions can then be expressed as formulas on the acquisition ofcompetencies or their components (competency elements) Descriptions specific toOpenCyc may use oc_CycLSentence-Assertible but simpler enumeration ofcompetency components could be used instead
Describing learning design rationales theories hypotheses guidelines and rulesThe concept of lsquodesignrsquo for our purposes is properly captured in the 5a definitionprovided in the Merriam Webster Online Dictionary (wwwm-wcom) lsquoan underlyingscheme that governs functioning developing or unfoldingrsquo In consequence learningdesigns refer to artefacts and not to the process of devising the educational solutionThis is equivalent to the concept oc_Design subsumed by the notion ofoc_Specification which is described as lsquoan abstract work that constitutes adescription of the properties of a Situation or a SomethingExisting and sometimeseven entire collections of such thingsrsquo In consequence the term LearningDesignwould simply delineate a subset of specifications that are oriented to describe howlearning experiences with concrete objectives should be Trivially oc_Designerdescribes the authors of the learning designs and oc_Designing represent theoc_HumanActivity that leads to the production of learning designs Designing as aspecial kind of activity is carried out (oc_doneBy) by a designer which provides a wayto attach metainformation about the rationale of the design
From an ontological perspective the departing problem in modelling learning designrationales is that their assumptions and hypotheses of different theoretical standpointsneed not be compatible or need not consider the same aspects of reality (Conole et al2004 Packer amp Goicoechea 2000) Furthermore different pedagogical approaches areknown to result in different mappings to activities (Conole et al 2004) This wouldeither require the provision of separate ontologies or the use of a representationalmechanism that allows such kind of potential inconsistency or divergence The conceptof microtheory in OpenCyc provides such representational mechanism intended to orga-nise assertions that depend on lsquoshared set of assumptions on which the truth of theassertions dependsrsquo Definitions inside the same microtheory need to be consistent butthis is not required across microtheories The descriptions provided in the earlier sectioncan be used as a base microtheory called LearningActivitiesMt for others that are spe-cific to concrete theoretical frameworks In this section we will describe elements thatare common to the description of theoretical frameworks the distinction among micro-theories will not be discussed
The intellectual processes of learning design in themselves will be represented asoc_Designing activities carried out by oc_Designers An additional consider-ation is that the information about the rationale of the design would be better attachedto representations of each of the design processes (ie to oc_Designing instances)than to the final design This enables the registering of different assumptions for differ-ent contributions to the same artefact eg different revisions from the same of differentdesigners
Semantic learning designs 341
copy 2006 The Author Journal compilation copy 2006 British Educational Communications and Technology Agency
Several designers may participate in the same designing activity and formal frameworksfor collaborative learning resource creation (Dodero Aedo amp Diacuteaz 2002) can be inte-grated by clearly specifying which class of things can be the outcome of a designingprocess For the sake of flexibility the concept of LearningDesignPart is introduced as anall-encompassing category that includes each of the elements described in the previoussection This includes not only activities and roles but also objectives conditions andthe like Furthermore the granularity of the descriptions allows for different levels fromconsidering a unique designing activity for the whole package to recording each ele-ment of the learning design comprehensively Because oc_Actions are subsumed bythe concept of oc_Event considered as a lsquodynamic situation in which the state of theworld changesrsquo each edition of a metadata record or any other form of description maybe subject to be described The concept is flexible enough to be able to record fine-grained detailed decisions This of course may include different versions of the sameartefacts but this is an orthogonal aspect of configuration that is not relevant to ourpresent discussion
Figure 2 depicts the main elements of the loci to which rationales assumptions andguidelines will be attached
A LearningDesigner can be defined as the class of humans that performactivities of learning resource design The consideration of learning design as aunique human activity could be considered as restrictive of intelligent softwarefor the task Simply changing the oc_Designer class to a subconcept ofoc_IntelligentAgent could remove this but this entails a change in Open-Cyc knowledge base The concept oc_Designing is a subclass of the conceptoc_ThinkingAboutTheStateOfTheWorld which represents lsquothinking pro-cessesrsquo which include mental events such as planning evaluating or inventingThus the concept of design described so far should be understood in terms of the
Figure 2 Main elements of the ontological schema for the loci of design rationales
oc_performedBy modifies
342 British Journal of Educational Technology Vol 37 No 3 2006
copy 2006 The Author Journal compilation copy 2006 British Educational Communications and Technology Agency
concept of lsquoexpanded rationalityrsquo as described by Hatchuel (2002) integrating cre-ativity and unexpected expansions of the original requirements This precludes onto-logical definitions in which the problem space is completely bounded a priori Inconsequence a degree of openness is necessary to integrate different kinds of detailin description from fully described ones to others with shallower semantics eg someproviding only references to generic assumptions
Theoretical frameworksWe will make the distinction that guidelines (as further defined) are the propositions thatdirectly predicate about the validity or appropriateness of learning design parts whileother theoretical propositions about learning that are not directly saying anythingabout the design of digital learning resources are considered as assumptions followingthe general sense provided in the Merriam-Webster dictionary lsquoa fact or statement (asa proposition axiom postulate or notion) taken for grantedrsquo
The fact that some assumption hypotheses or proposed law about learning is sustainedby a broader empirical base or is better sustained by rational means is out of the scopeof this work because here we focus on design practice (even though it is of an obviousinterest for researchers) In principle even unorthodox positions may be sustainedbecause the mechanism of microtheories and their inclusion mechanism allows forgreat flexibility in modelling them
Table 2 provides some examples of theoretical assumptions and related guidelines forlsquoextremersquo archetypical standpoints on learning These are not intended to reflect actualcommon positions but they are provided as illustrative examples of radical positionsthat could be modelled with our ontological schema
The examples provided in Table 2 go a step further than toolkits in Conole et al (2004)in making the design rationale explicit Nonetheless in many cases lsquocommon sensersquo isapplied to design and a shallower approach to specification could be used instead egproviding simply links to the theoretical standpoints used It should be noted also thatthe resulting guidelines vary greatly in their level of concreteness and some of themmay be considered as mandatory from some standpoints while others are mere generalrepresentations This would lead to different computational strategies
Guidelines and rulesThe WordNet (httpwordnetprincetonedu) thesaurus defines guidelines as lsquoguidancerelative to setting standards or determining a course of actionrsquo or lsquoa rule or principlethat provides guidance to appropriate behaviorrsquo In our case guidelines are statementsthat serve as criteria for decision in designing learning Given that e-learning designis still far from being a mature discipline guidelinesmdashas instances of recordedknowledgemdashcan be characterised as oc_NormativeSpecifications becausethey describe how learning resource design activities and their outcomes lsquoshould bersquoin a propositional form Propositions lsquohave some truth value in some context or
Semantic learning designs 343
copy 2006 The Author Journal compilation copy 2006 British Educational Communications and Technology Agency
ldquoworldrdquo rsquo Furthermore propositions lsquoare assumed to be representable (at least in prin-ciple) by a sentence in some formal or natural languagersquo This raises the need for alanguage that expresses guidelines in a general sense There are two main options forsuch representation be them formal logical sentences or reified guidelines
Formal logical sentence representation entails the formalisation through sentences ofthe constraints of the designs For example if lsquoevery learning resource must be activersquoaccording to some theoretical standpoint (eg one that fosters reflective learning) thissentence may be represented as an axiom and put into the appropriate separate micro-theory as the following CycL fragment
($forAll X ($implies
($isa X $LearningObject)
($resourceType X $Active))
Then resources that are not active are simply excluded from the notion itself This is astrong position in the sense that precludes some views on the learning design process
Table 2 Some examples of theoretical assumptions and guidelines
Archetypicalstandpoint Assumptions Guidelines
Instructivist(Behaviouristic)
bull Learning occurs through association and reinforcement
bull The results of learning can be appreciated through observable outcomes
(GB1) Every instance of oc-Learning entails theexposure to some LearningObjectthat produces the association (GB2) Learning Activitysequences should provide assessmentactivities that confirm theassociations that should have beenestablished so far
Constructivist bull Learning proceeds by building mental structures
(GC1) Agentrsquos mental structuresshould be represented(GC2) These structures should bepre-assessed in order to becomparable with the resulting ones(GC3) Design Activities arerecommended(GC4) Contrast Activities arerecommended
Sociocultural bull Learning occurs through social participatio
(GSC1) Every instance ofoc-Learning entails theinteraction of severaloc-IntelligentAgents asoc-Actors
344 British Journal of Educational Technology Vol 37 No 3 2006
copy 2006 The Author Journal compilation copy 2006 British Educational Communications and Technology Agency
Another typical example may be that of a sociocultural guideline that preclude Methodsthat do not entail multirole activities that provide access to social interaction ServicesThis approach to guidelines provides the best framework for automation and wouldenable the creation of microtheories that represent clearly formulated views onlearning Nonetheless there are two major problems for the use of this option as theunique mechanism for encoding design rationales
1 There is no clear-cut distinction between models and theories of learning becausethey come from a history of schools of thought with interrelations (Palmer 2001)This would lead to the need for the representation of dozens of variants reflectingpositions that only diverge in some aspects
2 Many guidelines cannot be represented as logical sentences because theyrepresent lsquorecommendationsrsquo or vague requirements eg see (GC3) or (GC4)(Table 2)
These issues suggest that a combination of formal sentences with reified guidelinesprovide the best option for flexibility The representation of guidelines has beenstudied in the field of HumanndashComputer-Interaction (HCI) in the last decade(Vanderdonckt 1999) and the main philosophy of such representations remainsvalid in the learning design domain The major challenge for such representations isthat of providing computational semantics that could be used for the implementationof software support but this is difficult to attain in all the cases Thus the concept oflearning design guideline will provide two kinds of representations general andschema-specific
The general use of instances of oc_NormativeSpecification allows the model-ling of lsquoconceptual works that describe how something should bersquo This may includerepresentation of known normative representations or procedures For example onemay refer to a concrete learning design model as that derived from Bannister (1981)which states that creativity cycles follow an initial phase of brainstorming (idea cap-ture) followed by a phase in which key issues are focused into an operation strategy(idea development) and a third phase in which project control takes place through alsquorecursive cyclersquo (operational management) The reference could be made by pointing tothe academic reference describing it as a conceptual work or it could be described asan Activity-Spec modelled explicitly inside the ontology As another examplesome detailed guidelines could be modelled by the normative specification sub-category of oc-Instructions defined as lsquoa sequence of tasks to be performed byan agentrsquo
Schema-specific guidelines restrict the freedom to represent criteria as conceptual workinstances to a concrete and commonly agreed ontology that determine what can be saidabout how learning designs should be according to a theoretical standpoint or frame-work These build on an activityndashrole ontology as the one described in the previoussection and add a prescription on the kind of lsquoshould-bersquo predications that are allow-able Table 3 describes examples of a minimal language that could be used for thatpurpose
Semantic learning designs 345
copy 2006 The Author Journal compilation copy 2006 British Educational Communications and Technology Agency
In the examples CAT represents a category of individual expressed through a conceptor a logical formula and the sentence is a synthetic way of expressing some recom-mended or mandatory constraint over such category Other languages could be engi-neered in further research but the one described can be directly expressed in OpenCycand combined with software modules that check or enforce them in some way Notethat this representation does not use direct OpenCyc sentences so that reification is usedinstead of direct logical sentences It is not the aim of this paper to determine the mostpowerful and flexible representation language but only set a foundation for furtherspecification efforts
Analysis and comparison of two theoretical semantic frameworksAs an assessment of the representational capabilities of the scheme just described acase study of representation is sketched in this section We will use the learning designtoolkit described by Conole et al (2004) as the underlying terminology even thoughother frameworks could be used instead
Two rather diverging positions for design will be used as an illustration Table 4 sum-marises the two design scenarios
Table 3 Possible language for use in learning designs
Sentence type Potential uses Example
RecommendationshouldBe (CATSENT)
Recommend kinds ofactivities
Design activities are recommended (GC3)shouldBe(Method subEvent (CATDesignActivity))
ObligationmustBe (CATSENT)
Some activity typesmust follow others
Assessments must follow each expositionto contents (GD1) mustBe(and(ExpositiveActivityIndividualActivity)startAfterEndingOf(CAT
AssessmentActivity)
Activities must havesome characteristic
Every activity must entail interactionmustBe(LDActivity(usedIn(CAT
InteractiveService)))Every activity must provide acontent-oriented learning object forindividual use (similar to GB1)mustBe(LDActivity(and (usedIn(CAT
LearningObject)(learnerRoleNumber(1))))
Some structure shouldbe enforced in methods
Methods should have a sequential actstructuremustBe(Methodnot ((subEvent(CAT Play))))
CAT = category GC3 = Constructivist Guideline 3 (see Table 2) GD1 = Guideline 1
346 British Journal of Educational Technology Vol 37 No 3 2006
copy 2006 The Author Journal compilation copy 2006 British Educational Communications and Technology Agency
The following subsections sketch the main outcomes of an example learning designprocess for both scenarios assuming that the process described by Conole et al (2004)is used in both situations
Differences in modelling the design processThe forces driving the design process in both scenarios are fairly diverging The designof Scenario A will typically be constrained by a clear target specified in terms of anaggregated competency level while in Scenario B there are no concrete competenciesto be acquired and the main objective of the design activity would be twofold On theone hand the activities are targeted at promoting interchange of ideas in an attemptto stimulate casual convergence of interest which is the main value of the culture ofthe special interest group On the other hand the activities have the generic objectiveof strengthening social ties which is an important element of cohesion for the under-taking of joint projects
Table 5 describes the main inputs and possible activities and their objectives at a highlevel
The changes intended to be achieved in each of the scenarios are stated differently witha formal definition in (A) and a shallow cognitive and social representation in the case
Table 4 Learning design scenarios
Scenario DescriptionClassification according to
Conole et al (2004) toolkit
(A) Systemic-Behaviourist A software development company attempts to designlearning units about newtechnologies or softwareproducts for its developerstaff The design is drivenby a cycle of lsquocompetencygap analysisrsquo that modelsorganisational learning in terms of the aggregated competencies of the employeesmdashsimilar to theone described by Sicilia(2005)
Focuses on the individual where a body of informationforms the basis and rawmaterial for learning Learning is mainly considered as nonreflective ie based on skill learning and memorisation
(B) Sociocultural A special interest group ona concrete research topicaims at developing the knowledge of its membersby sharing experiencesand discussing on themesof common interest
Focuses on the social whereexperience is the main source of learning Conscious reflection on experience is transformed into learning
Semantic learning designs 347
copy 2006 The Author Journal compilation copy 2006 British Educational Communications and Technology Agency
of (B) In addition case (B) requires a model of social network that is largely irrelevantin case (A)
Differences in recording activity rationalesThe approach for the recording of design rationales also differs because of the differentcontexts Table 6 summarises the main elements for each of the scenarios
Conclusions and directions for researchCurrent technology and practice related to the description of reusable digital learningresources has resulted in techniques that allow the description of learning activities of
Table 5 Inputs activities and objectives of two learning design scenarios
Scenario Design inputs Activities and their objectives
(A) Systemic-Behaviourist A formal competency recordfor employees and competencyrequirements expressed in terms of amounts of required competencies (assuming a shared measurement scale) Such requirements are actually derived from strategy or from the contingent requirements of upcoming projects
Competency instances are used to measure the levelof Competency-Spec for the required competencies (or their constituents) A formal match considering the agenda of employees is carried out for determining the target learners LearningObjects forindividual use are selected asthe main elements of theactivities and the criteria forselection are again a match ofthe objectives stated inmetadata (as stated incategory 9 of Learning Object MetadatamdashClassification) and thecompetencies required
(B) Sociocultural The expertise of the special interest group members as stated in the concepts addressed by their publicationsprojects or other activities
Informal group activities are the main approach applicablesince there is not a predefinedgroup structure but a cultureof sharing and socialisation The results of the sharing of ideas can be expressed in terms of oc-knowsAboutpredicates and social ties will be modelled through positive valence FOAF-knows(httpxmlnscomfoaf01)predicates with an added numerical value representingstrength
348 British Journal of Educational Technology Vol 37 No 3 2006
copy 2006 The Author Journal compilation copy 2006 British Educational Communications and Technology Agency
a complex kind including their intended participants educational objectives and exe-cution constraints The representation and recording of the assumptions and guidelinesthat were considered in the creation of such resources would extend current techniquesto capture the rationale of the design itself This in turn will eventually result in theavailability of a large base of empirical data available in which theoretical assumptionsor rules are explicitly connected to the resulting designs as these could be linked withthe actual results and trace of the learning processes that used them The resulting bodyof evidence would represent a critical asset in the development of the theory of learningmediated through computers
This paper has described the first attempt to extend existing learning technology modelswith a schema explicitly targeted at describing learning design rationales The OpenCycknowledge base has been used as the semantic reference from which the ontologicalschema has been specified Concretely an activity-based schema based on IMS LD hasbeen described and theoretical assumptions and codified guidelines have been intro-duced The flexibility of the representation schema has been analysed through therepresentation of two stereotyped theoretical positions
Two main research directions should continue the work described here On the onehand further ontological engineering effort is required to come up with a more com-prehensive and rich schema for learning design rationales On the other hand tool
Table 6 Main elements of the two learning design scenarios
Scenario Approach Examples
(A) Systemic-Behaviourist Formal logical sentencesare used to guarantee that every design followsan organisationalobjective
It is required that the objective of every Method is attached to a Competency-Spec instance as objective
Formal assessment is a requirement foraccountability purposes
Guideline (GD1) is attached to everyDesigningLearningResource instance that produces an Act
(B) Sociocultural The interaction in informal settings can beused to derive social models (Sicilia amp Garciacutea2004)
The interactions in discussion canbe used to measure the degree of lsquoproximityrsquo between members Therationale of the activities can bejustified in terms of SocialTies inside SocialNetworks Further the focus of each groupactivity should be clearly stated anda combination of the increase in collective knowledge and the strengthening of social relations can be used for assessment
Semantic learning designs 349
copy 2006 The Author Journal compilation copy 2006 British Educational Communications and Technology Agency
implementation and record gathering efforts would be required to attain the ultimateobjective of the ontological definitions described so far With such effort an unprece-dented research material could be available in some years providing improved oppor-tunities for scientific inquiry on design as an activity regarding pedagogy
Other potential directions for research may be the description of interpretations of thedesign intentions (for which no rationale was explicitly recorded) the mining or detec-tion of patterns as related to rationale description or the detailed use of theoreticalrationales in agents that support the learning process which represent a category ofrelevant learning support software (Holmes in press)
ReferencesAllert H (2004) Coherent social systems for learningmdashan approach for contextualized and
community-centred metadata Journal of Interactive Media in Education 2004 1Baader F Calvanese D McGuinness D Nardi D amp Patel-Schneider P (Eds) (2003) The
description logic handbook Theory implementation and applications Cambridge CambridgeUniversity Press
Bannister D (1981) Personal construct theory and research method In P Reason amp J Rowan(Eds) Human inquiry a sourcebook of new paradigm research (pp 191ndash200) Chichester UKJohn Wiley and Sons
Berners-Lee T Hendler J amp Lassila O (2001) The semantic web Scientific American 284 534ndash43
Brown A amp Voltz B (2005) Elements of effective e-learning design International Review ofResearch in Open and Distance Learning 6 1
Buzza D Bean D Harrigan K amp Carey T (2005) Learning design repositories adaptinglearning design specifications for shared instructional knowledge Canadian Journal of Learningand Technology 30 3
Conole G Dyke M Oliver M amp Seale J (2004) Mapping pedagogy and tools for effectivelearning design Computers amp Education 43 1ndash2 17ndash33
Dodero J M Aedo I amp Diacuteaz P (2002) Participative knowledge production of learning objectsfor e-books The Electronic Library 20 4 296ndash305
Friesen N (2005) Interoperability and learning objects an overview of e-learning standardiza-tion Interdisciplinary Journal of Knowledge and Learning Objects 1 1 23ndash31
Gruber T (1995) Towards principles for the design of ontologies used for knowledge sharingInternational Journal of HumanndashComputer Studies 43 56 907ndash928
Hatchuel A (2002) Towards design theory and expandable rationality the unfinished programof Herbert Simon Journal of Management and Governance 5 3ndash4 260ndash271
Holmes J (in press) Designing agents to support learning by explaining Computers and EducationIEEE Learning Technology Standards Committee (2002) Learning object metadata (LOM) Final
Draft Standard IEEE 1484121ndash2002IMS (2003) IMS Learning Design Information Model (Version 10) Final Specification 20
January 2003Kaptelinin V Kuutti K amp Bannon L (1995) Activity theory basic concepts and applications
In Blumenthalet al (Eds) Humanndashcomputer interaction Lecture Notes in Computer Science 1015189ndash201
Koper R (2004) Use of the semantic web to solve some basic problems in education increaseflexible distributed lifelong learning decrease teacherrsquos workload Journal of Interactive Mediain Education 2004 6
Lenat D (1995) CYC a large-scale investment in knowledge infrastructure Communications ofthe ACM 38 11 33ndash38
350 British Journal of Educational Technology Vol 37 No 3 2006
copy 2006 The Author Journal compilation copy 2006 British Educational Communications and Technology Agency
McGreal R (2004) Learning objects a practical definition International Journal of InstructionalTechnology and Distance Learning 1 9 21ndash32
Packer M amp Goicoechea J (2000) Sociocultural and constructivist theories of learningontology not just epistemology Educational Psychologist 35 4 227ndash241
Palmer J (2001) Fifty major thinkers on education London and New York RoutledgePolsani P R (2003) Use and abuse of reusable learning objects Journal of Digital Information 3
4 Rothwell W amp Kazanas H (1992) Mastering the instructional design process San Francisco CA
Jossey-BassSaacutenchez-Alonso S amp Sicilia M A (2004) Relationships and commitments in learning object
metadata In Proceedings of the 5th International Conference on Information Technology BasedHigher Education and Training ITHET 2004 Istanbul Turkey
Sicilia M A (2005) Ontology-Based Competency Management Infrastructures for theKnowledge-intensive Learning Organization In Lytras amp Naeve (Eds) Intelligent Learning Infra-structures in Knowledge Intensive Organizations A Semantic Web perspective IDEA USA pp 302ndash324
Sicilia M A amp Garciacutea E (2004) Fuzzy group models for adaptation in cooperative informationretrieval contexts Lecture Notes in Computer Science 2932 324ndash334
Sicilia M A amp Garciacutea E (2005) On the convergence of formal ontologies and standardizede-learning Journal of Distance Education Technologies 3 2 13ndash29
Sicilia M A amp Lytras M (2005) On the representation of change according to different ontol-ogies of learning International Journal of Learning and Change 1 1 66ndash79
Sicilia M A Lytras M Rodriacuteguez E amp Garciacutea E (in press) Integrating descriptions of knowl-edge management learning activities into large ontological structures a case study Data andKnowledge Engineering
Tennant M (1991) Expertise as a dimension of adult development New Education 13 1 49ndash55Vanderdonckt J (1999) Development milestones towards a tool for working with guidelines
Interacting with Computers 12 2 81ndash118Wand Y amp Weber R (1995) On the deep structure of information systems Information Systems
Journal 5 203ndash223Wiley D A (2001) The instructional use of learning objects Bloomington IN Association for
Educational Communications and Technology
Semantic learning designs
335
copy 2006 The Author Journal compilation copy 2006 British Educational Communications and Technology Agency
Describing the main structural elements of IMS LD
The conceptual model of IMS LD (IMS 2003) provides a concise lsquometalanguagersquo thatcould capture a diversity of pedagogical approaches In doing so it essentially adheresto an activity-centred paradigm (Allert 2004) It should be noted that this model isneutral to specific learning theories Even though it integrates the hierarchical struc-ture of activities and mediation principles of
Activity Theory
(Kaptelinin Kuutti amp Ban-non 1995) it is actually a metalanguage that does not make assumptions about otherelements of such theory as the internalisationexternalisation differentiation
The first step in developing an ontological schema for IMS LD must be that of clarifyingthe main entities to be represented
OpenCyc
provides a rich framework to describetemporal activities concretely the
oc_PurposefulAction
concept (a kind of
oc_Event
) defined as lsquoan action consciously volitionally and purposefully done by atleast one actorrsquo captures the essence of learning activities which are the ultimate targetof the specification Nonetheless IMS LD metadata does not attempt to representing theactivities as actually enacted by persons but they describe
blueprints
that will eventuallybe bound to one or several instantiations of activities and they are intended to besearched as such (Buzza Bean Harrigan amp Carey 2005) In contrast
oc_Event
s andtheir associated entities are related to instantiations and not to their specificationsFurther the concept of
oc_ActorSlot
represents the specific roles played by personsin events so that this could be used to represent the LD
staff
and
learner
role categorybut they do not represent roles understood as templates for their eventual binding toconcrete people
An alternative to representing LD specifications as blueprints is that of creating
Event
types for each activity defined in a learning design But this approach entails at leasttwo problems On the one hand this is not ontologically correct because it does notrepresent the reality that the production of IMS LD is an act of creating specificationartefacts instead of a process of classification of possible real world entities In factmany learning designs will probably be never executed and some of them may bepurposefully created as mere design examples On the other hand the direct creationof learning designs as ontology terms entails that the activities actually performed mustadhere strictly to the original learning design used which severely restricts the flexibil-ity in using the design with ad hoc changes or modifications (or even modified
versions
)This second practical flaw suggests that LD specifications should be represented asseparated entities
This has led us to the major decision of modelling learning designs as specificationsfollowing the definition of
oc_Specification
as lsquoan abstract work that constitutesa description of the properties of a
Situation
or a
SomethingExisting
and sometimes evenentire collections of such thingsrsquo Table 1 summarises the representation of
LearningDesigns
in terms of specifications (specialisations of
oc_Specification
) their mainrelated properties and the implications of each of the specifications when producingactual activities Notifications and completion conditions are omitted for brevityThe concepts of lsquo
learning design
rsquo and lsquo
method
rsquo can be considered as equivalent from an
336
British Journal of Educational Technology Vol 37 No 3 2006
copy 2006 The Author Journal compilation copy 2006 British Educational Communications and Technology Agency
Table 1 Summary of the main specification definitions in the ontological schema
Concept Main descriptions Consequences in actual activities
Method-Spec Objectives and prerequisites(described in the followingsection) Structured in
play
specifications or directly
act
specifications
Instantiation is an independentmeaningful learning activity Playsare cotemporal
oc_subEvent
ssubsumed in the method Note thatthe semantic definition of subeventsyet entails the subsuming of subeventsin the containing event
Play-Spec Collection of sequential acts thatcomprise the play
Because plays are potentiallyconcurrent there are no furthertemporal restrictions
Act-Spec Collection of concurrent role parts that detail the activitiesincluded in the design Role partsand environments are the mainelements related to acts
The sequence of instantiations of actscould be sequenced through the
oc_startsAfterEndingOf
RolendashPart-Spec The linking of role specificationsto concrete activities or activitystructures
This entity is not required in theontological schema because theparticipation of roles in the activitiescan be stated in terms of specificationsof
oc_ActorRole
s or subeventsthat are not of a temporal nature
Activity-Spec The specification of learning andsupport structures in terms of
Role-Spec
s
ibtUsed LearningObject
s and used
Service
s Concrete objectivescan also be attached
Each instantiation of an
Activity-Spec
will result ininstances of specific types of
oc_Activity
Following LD specs
support
activities instantiate for everyrole mapping while
learning
onesinstantiate for all the role mappings
Environment These are descriptions of learning objects and services that do not require anexplicit mapping in the ontological schema
Service Service facilities have to belsquoinstantiated by a local run-timeservicesrsquo eg a chat or discussionforum Their computationalnature can be described interms of concrete
oc_ComputerProgramCW
Subcategories of such conceptualrepresentations of softwareentities can be used for thetypology of services
The
oc_ComputerProcessRunning
term is used to describe executionsof
oc_ComputerProgramCW
which cover every instance of aservice as understoodin IMS LD Thelinking between conceptual andactual running instances is coveredby the required predicate
oc-programOfProcess
Semantic learning designs
337
copy 2006 The Author Journal compilation copy 2006 British Educational Communications and Technology Agency
Learning object Those descriptions provided bylearning object metadataLearning objects can be definedas
oc_IBT
s as described inSicilia Lytras Rodriacuteguez andGarciacutea (in press) and approachesto describing metadata throughontologies yet existmdashsee an overview in Sicilia and Garciacutea(2005)
The learning objects selected arerepresented as instances in theontology which are declared to beused in activities through the
ibtUsed
predicate
Role-Spec Describe stereotypical actors oflearning and support activitiesThese act as lsquocontractrsquo for theactual participants in the senseprovided by Saacutenchez-Alonso andSicilia (2004)
Role specs are mapped to actualusers when the activities take placeThis is mapped to predicates of theclass
oc_ActorRole
Concept Main descriptions Consequences in actual activities
LD
=
Learning Design
Table 1 Continued
ontological perspective as the former allow onemdashand only onemdashinstance of the latterwith no particular logical differences
As described in Table 1 the
OpenCyc
built-in support for modelling temporal events canbe used to express the coordination relationships between the constituents of the unitsof learning The actual instances of learning events will be modelled as instances of
oc_Learning
(lsquothe collection of all events brief or extended in which an agent isacquiring information or know-howrsquo) and more specifically as instances of a speciali-sation
ComputerMediatedLearning
simply requiring the use of information net-works for at least part of the learning activities Because events in
OpenCyc
are alsosituations they have components in addition to the temporal extent including peopleinvolved information things used and other arrangements
Figure 1b depicts the main elements that describe actual current or past enactmentsof activities The schema is based on a clear separation of the specifications and theactual activities that by itself constitutes a metadata layering
OpenCyc
predicatesas
oc_situationConformsToSpecification
allows the linking of situations(every action is a situation also) to the specifications they conformmdashbe it purposely ornot The
LD-Activity term has been introduced as the common category of learningand support activities Methods plays and acts become special subcategories of LD-Activity and the constituent activities of Acts are instances of LD Activity also Rolesare attached to any kind of LD Activity even though it is required for the low-levelstructure The learning design concept of lsquoactivity structurersquo as an aggregation ofactivities does not require an explicit modelling because of the generic mereology ofactivities engineered in OpenCyc
338 British Journal of Educational Technology Vol 37 No 3 2006
copy 2006 The Author Journal compilation copy 2006 British Educational Communications and Technology Agency
Figure 1 (a) Summary of the competency components in the ontological schema (b) Summary of the main activityndashrole definitions in the ontological schema
oc_WorkingEvent
oc_IntelligentAgent
The competency elements aremeasurable characteristics andthey will be typed by the relatedSpecification classes eg Skill
SpecputlnPracticeln
(a)
Competency Competency-Spec
Competencyelement
Skill oc_ Competence
Attitude
Knowledge
has
thingSpecified
oc_Learning
ComputerMediatedLearning
oc_Role
(b)
Staff
Learner
oc_Specification
Method-Spec
Play-Spec
Act-Spec
Support
LD-Activity
contains
contains
Method
Play
Act
startAfterEndingOf
subAction
subAction
Activity-Spec
ltltActorSlotsgtgt
situationConformsToSpec
Activity-Spec
Activity
Semantic learning designs 339
copy 2006 The Author Journal compilation copy 2006 British Educational Communications and Technology Agency
The concept of oc_Role accounts not only for participant roles but also for otherkinds of lsquocomponentsrsquo This is why two specialisations of the concept ofoc_ActorSlot labelled Staff and Learner are required to differentiate partici-pants from other elements As is obvious from the earlier discussion the built-in rep-resentation of activities in OpenCyc is powerful enough for the definition of learningdesign activities
The final part of the ontological specification described so far is that of the mappingfrom LD specifications to the actual past or present activity realisations Theoc_situationConformsToSpecification predicate can be used to trace whichactivity realizations come from a given learning design activity definition and the sameoccurs with the oc_thingSpecified for other learning design elements such asroles or environments Existing learning design execution engines as CopperCore(httpcoppercoreorg) actually map LD specifications to online collaboration so thatthe translation of the ontological representations of learning designs to ontology defi-nitions of activities is straightforward following similar representation
Describing objectives outcomes and preconditionsAccording to IMS learning design learning objectives and prerequisites lsquocan bedescribed using the IMS Reusable Definition of Competency or Educational Objective(RDCEO) format but can also refer to simple resources (eg a text) with a descriptionof the learning objectiversquo As the latter option is not useful from the viewpoint of com-putational semantics we will adhere to the former As commented earlier the compe-tency view is not exclusive to other kinds of objectives related to other ontologicalstandpoints (Sicilia amp Lytras 2005) and other conceptions of change or improvementmay include social ties cognitive structures or the combination of the knowledge of agroup In fact competencies adhere mostly to behaviouristic frameworks that focus onobservable external behaviour (Tennant 1991) rather than on cognitive representa-tions Nonetheless the further discussion is equally applicable to other kinds of objec-tives so that we will adhere to the competency view
oc_Competence in OpenCyc is defined as a quantity that lsquofor describing the level ofskill with which an agent performs some taskrsquo This does not remove the lack of preci-sion that is yet present in RDCEO definitions as described by Sicilia (2005) As analternative Figure 1a provides a simplified account of the model described by Sicilia asrelated to OpenCyc terms in which competence is integrated as one of the elements of thebroader notion of competency according to the model of Rummel (Rothwell amp Kazanas1992) which considers work situations attributes of the individual that are applied insuch situations and measurement with respect to some standard of performance Themost relevant distinction is again the separation of actual competency representationsand competency lsquodefinitionsrsquo (ie specifications) intended to provide a type structure tocompetencies to be used in objective or precondition specification In this case thegeneric oc-thingSpecified predicate can be used for the linking of both levels ofrepresentation
340 British Journal of Educational Technology Vol 37 No 3 2006
copy 2006 The Author Journal compilation copy 2006 British Educational Communications and Technology Agency
Objectives and preconditions can then be expressed as formulas on the acquisition ofcompetencies or their components (competency elements) Descriptions specific toOpenCyc may use oc_CycLSentence-Assertible but simpler enumeration ofcompetency components could be used instead
Describing learning design rationales theories hypotheses guidelines and rulesThe concept of lsquodesignrsquo for our purposes is properly captured in the 5a definitionprovided in the Merriam Webster Online Dictionary (wwwm-wcom) lsquoan underlyingscheme that governs functioning developing or unfoldingrsquo In consequence learningdesigns refer to artefacts and not to the process of devising the educational solutionThis is equivalent to the concept oc_Design subsumed by the notion ofoc_Specification which is described as lsquoan abstract work that constitutes adescription of the properties of a Situation or a SomethingExisting and sometimeseven entire collections of such thingsrsquo In consequence the term LearningDesignwould simply delineate a subset of specifications that are oriented to describe howlearning experiences with concrete objectives should be Trivially oc_Designerdescribes the authors of the learning designs and oc_Designing represent theoc_HumanActivity that leads to the production of learning designs Designing as aspecial kind of activity is carried out (oc_doneBy) by a designer which provides a wayto attach metainformation about the rationale of the design
From an ontological perspective the departing problem in modelling learning designrationales is that their assumptions and hypotheses of different theoretical standpointsneed not be compatible or need not consider the same aspects of reality (Conole et al2004 Packer amp Goicoechea 2000) Furthermore different pedagogical approaches areknown to result in different mappings to activities (Conole et al 2004) This wouldeither require the provision of separate ontologies or the use of a representationalmechanism that allows such kind of potential inconsistency or divergence The conceptof microtheory in OpenCyc provides such representational mechanism intended to orga-nise assertions that depend on lsquoshared set of assumptions on which the truth of theassertions dependsrsquo Definitions inside the same microtheory need to be consistent butthis is not required across microtheories The descriptions provided in the earlier sectioncan be used as a base microtheory called LearningActivitiesMt for others that are spe-cific to concrete theoretical frameworks In this section we will describe elements thatare common to the description of theoretical frameworks the distinction among micro-theories will not be discussed
The intellectual processes of learning design in themselves will be represented asoc_Designing activities carried out by oc_Designers An additional consider-ation is that the information about the rationale of the design would be better attachedto representations of each of the design processes (ie to oc_Designing instances)than to the final design This enables the registering of different assumptions for differ-ent contributions to the same artefact eg different revisions from the same of differentdesigners
Semantic learning designs 341
copy 2006 The Author Journal compilation copy 2006 British Educational Communications and Technology Agency
Several designers may participate in the same designing activity and formal frameworksfor collaborative learning resource creation (Dodero Aedo amp Diacuteaz 2002) can be inte-grated by clearly specifying which class of things can be the outcome of a designingprocess For the sake of flexibility the concept of LearningDesignPart is introduced as anall-encompassing category that includes each of the elements described in the previoussection This includes not only activities and roles but also objectives conditions andthe like Furthermore the granularity of the descriptions allows for different levels fromconsidering a unique designing activity for the whole package to recording each ele-ment of the learning design comprehensively Because oc_Actions are subsumed bythe concept of oc_Event considered as a lsquodynamic situation in which the state of theworld changesrsquo each edition of a metadata record or any other form of description maybe subject to be described The concept is flexible enough to be able to record fine-grained detailed decisions This of course may include different versions of the sameartefacts but this is an orthogonal aspect of configuration that is not relevant to ourpresent discussion
Figure 2 depicts the main elements of the loci to which rationales assumptions andguidelines will be attached
A LearningDesigner can be defined as the class of humans that performactivities of learning resource design The consideration of learning design as aunique human activity could be considered as restrictive of intelligent softwarefor the task Simply changing the oc_Designer class to a subconcept ofoc_IntelligentAgent could remove this but this entails a change in Open-Cyc knowledge base The concept oc_Designing is a subclass of the conceptoc_ThinkingAboutTheStateOfTheWorld which represents lsquothinking pro-cessesrsquo which include mental events such as planning evaluating or inventingThus the concept of design described so far should be understood in terms of the
Figure 2 Main elements of the ontological schema for the loci of design rationales
oc_performedBy modifies
342 British Journal of Educational Technology Vol 37 No 3 2006
copy 2006 The Author Journal compilation copy 2006 British Educational Communications and Technology Agency
concept of lsquoexpanded rationalityrsquo as described by Hatchuel (2002) integrating cre-ativity and unexpected expansions of the original requirements This precludes onto-logical definitions in which the problem space is completely bounded a priori Inconsequence a degree of openness is necessary to integrate different kinds of detailin description from fully described ones to others with shallower semantics eg someproviding only references to generic assumptions
Theoretical frameworksWe will make the distinction that guidelines (as further defined) are the propositions thatdirectly predicate about the validity or appropriateness of learning design parts whileother theoretical propositions about learning that are not directly saying anythingabout the design of digital learning resources are considered as assumptions followingthe general sense provided in the Merriam-Webster dictionary lsquoa fact or statement (asa proposition axiom postulate or notion) taken for grantedrsquo
The fact that some assumption hypotheses or proposed law about learning is sustainedby a broader empirical base or is better sustained by rational means is out of the scopeof this work because here we focus on design practice (even though it is of an obviousinterest for researchers) In principle even unorthodox positions may be sustainedbecause the mechanism of microtheories and their inclusion mechanism allows forgreat flexibility in modelling them
Table 2 provides some examples of theoretical assumptions and related guidelines forlsquoextremersquo archetypical standpoints on learning These are not intended to reflect actualcommon positions but they are provided as illustrative examples of radical positionsthat could be modelled with our ontological schema
The examples provided in Table 2 go a step further than toolkits in Conole et al (2004)in making the design rationale explicit Nonetheless in many cases lsquocommon sensersquo isapplied to design and a shallower approach to specification could be used instead egproviding simply links to the theoretical standpoints used It should be noted also thatthe resulting guidelines vary greatly in their level of concreteness and some of themmay be considered as mandatory from some standpoints while others are mere generalrepresentations This would lead to different computational strategies
Guidelines and rulesThe WordNet (httpwordnetprincetonedu) thesaurus defines guidelines as lsquoguidancerelative to setting standards or determining a course of actionrsquo or lsquoa rule or principlethat provides guidance to appropriate behaviorrsquo In our case guidelines are statementsthat serve as criteria for decision in designing learning Given that e-learning designis still far from being a mature discipline guidelinesmdashas instances of recordedknowledgemdashcan be characterised as oc_NormativeSpecifications becausethey describe how learning resource design activities and their outcomes lsquoshould bersquoin a propositional form Propositions lsquohave some truth value in some context or
Semantic learning designs 343
copy 2006 The Author Journal compilation copy 2006 British Educational Communications and Technology Agency
ldquoworldrdquo rsquo Furthermore propositions lsquoare assumed to be representable (at least in prin-ciple) by a sentence in some formal or natural languagersquo This raises the need for alanguage that expresses guidelines in a general sense There are two main options forsuch representation be them formal logical sentences or reified guidelines
Formal logical sentence representation entails the formalisation through sentences ofthe constraints of the designs For example if lsquoevery learning resource must be activersquoaccording to some theoretical standpoint (eg one that fosters reflective learning) thissentence may be represented as an axiom and put into the appropriate separate micro-theory as the following CycL fragment
($forAll X ($implies
($isa X $LearningObject)
($resourceType X $Active))
Then resources that are not active are simply excluded from the notion itself This is astrong position in the sense that precludes some views on the learning design process
Table 2 Some examples of theoretical assumptions and guidelines
Archetypicalstandpoint Assumptions Guidelines
Instructivist(Behaviouristic)
bull Learning occurs through association and reinforcement
bull The results of learning can be appreciated through observable outcomes
(GB1) Every instance of oc-Learning entails theexposure to some LearningObjectthat produces the association (GB2) Learning Activitysequences should provide assessmentactivities that confirm theassociations that should have beenestablished so far
Constructivist bull Learning proceeds by building mental structures
(GC1) Agentrsquos mental structuresshould be represented(GC2) These structures should bepre-assessed in order to becomparable with the resulting ones(GC3) Design Activities arerecommended(GC4) Contrast Activities arerecommended
Sociocultural bull Learning occurs through social participatio
(GSC1) Every instance ofoc-Learning entails theinteraction of severaloc-IntelligentAgents asoc-Actors
344 British Journal of Educational Technology Vol 37 No 3 2006
copy 2006 The Author Journal compilation copy 2006 British Educational Communications and Technology Agency
Another typical example may be that of a sociocultural guideline that preclude Methodsthat do not entail multirole activities that provide access to social interaction ServicesThis approach to guidelines provides the best framework for automation and wouldenable the creation of microtheories that represent clearly formulated views onlearning Nonetheless there are two major problems for the use of this option as theunique mechanism for encoding design rationales
1 There is no clear-cut distinction between models and theories of learning becausethey come from a history of schools of thought with interrelations (Palmer 2001)This would lead to the need for the representation of dozens of variants reflectingpositions that only diverge in some aspects
2 Many guidelines cannot be represented as logical sentences because theyrepresent lsquorecommendationsrsquo or vague requirements eg see (GC3) or (GC4)(Table 2)
These issues suggest that a combination of formal sentences with reified guidelinesprovide the best option for flexibility The representation of guidelines has beenstudied in the field of HumanndashComputer-Interaction (HCI) in the last decade(Vanderdonckt 1999) and the main philosophy of such representations remainsvalid in the learning design domain The major challenge for such representations isthat of providing computational semantics that could be used for the implementationof software support but this is difficult to attain in all the cases Thus the concept oflearning design guideline will provide two kinds of representations general andschema-specific
The general use of instances of oc_NormativeSpecification allows the model-ling of lsquoconceptual works that describe how something should bersquo This may includerepresentation of known normative representations or procedures For example onemay refer to a concrete learning design model as that derived from Bannister (1981)which states that creativity cycles follow an initial phase of brainstorming (idea cap-ture) followed by a phase in which key issues are focused into an operation strategy(idea development) and a third phase in which project control takes place through alsquorecursive cyclersquo (operational management) The reference could be made by pointing tothe academic reference describing it as a conceptual work or it could be described asan Activity-Spec modelled explicitly inside the ontology As another examplesome detailed guidelines could be modelled by the normative specification sub-category of oc-Instructions defined as lsquoa sequence of tasks to be performed byan agentrsquo
Schema-specific guidelines restrict the freedom to represent criteria as conceptual workinstances to a concrete and commonly agreed ontology that determine what can be saidabout how learning designs should be according to a theoretical standpoint or frame-work These build on an activityndashrole ontology as the one described in the previoussection and add a prescription on the kind of lsquoshould-bersquo predications that are allow-able Table 3 describes examples of a minimal language that could be used for thatpurpose
Semantic learning designs 345
copy 2006 The Author Journal compilation copy 2006 British Educational Communications and Technology Agency
In the examples CAT represents a category of individual expressed through a conceptor a logical formula and the sentence is a synthetic way of expressing some recom-mended or mandatory constraint over such category Other languages could be engi-neered in further research but the one described can be directly expressed in OpenCycand combined with software modules that check or enforce them in some way Notethat this representation does not use direct OpenCyc sentences so that reification is usedinstead of direct logical sentences It is not the aim of this paper to determine the mostpowerful and flexible representation language but only set a foundation for furtherspecification efforts
Analysis and comparison of two theoretical semantic frameworksAs an assessment of the representational capabilities of the scheme just described acase study of representation is sketched in this section We will use the learning designtoolkit described by Conole et al (2004) as the underlying terminology even thoughother frameworks could be used instead
Two rather diverging positions for design will be used as an illustration Table 4 sum-marises the two design scenarios
Table 3 Possible language for use in learning designs
Sentence type Potential uses Example
RecommendationshouldBe (CATSENT)
Recommend kinds ofactivities
Design activities are recommended (GC3)shouldBe(Method subEvent (CATDesignActivity))
ObligationmustBe (CATSENT)
Some activity typesmust follow others
Assessments must follow each expositionto contents (GD1) mustBe(and(ExpositiveActivityIndividualActivity)startAfterEndingOf(CAT
AssessmentActivity)
Activities must havesome characteristic
Every activity must entail interactionmustBe(LDActivity(usedIn(CAT
InteractiveService)))Every activity must provide acontent-oriented learning object forindividual use (similar to GB1)mustBe(LDActivity(and (usedIn(CAT
LearningObject)(learnerRoleNumber(1))))
Some structure shouldbe enforced in methods
Methods should have a sequential actstructuremustBe(Methodnot ((subEvent(CAT Play))))
CAT = category GC3 = Constructivist Guideline 3 (see Table 2) GD1 = Guideline 1
346 British Journal of Educational Technology Vol 37 No 3 2006
copy 2006 The Author Journal compilation copy 2006 British Educational Communications and Technology Agency
The following subsections sketch the main outcomes of an example learning designprocess for both scenarios assuming that the process described by Conole et al (2004)is used in both situations
Differences in modelling the design processThe forces driving the design process in both scenarios are fairly diverging The designof Scenario A will typically be constrained by a clear target specified in terms of anaggregated competency level while in Scenario B there are no concrete competenciesto be acquired and the main objective of the design activity would be twofold On theone hand the activities are targeted at promoting interchange of ideas in an attemptto stimulate casual convergence of interest which is the main value of the culture ofthe special interest group On the other hand the activities have the generic objectiveof strengthening social ties which is an important element of cohesion for the under-taking of joint projects
Table 5 describes the main inputs and possible activities and their objectives at a highlevel
The changes intended to be achieved in each of the scenarios are stated differently witha formal definition in (A) and a shallow cognitive and social representation in the case
Table 4 Learning design scenarios
Scenario DescriptionClassification according to
Conole et al (2004) toolkit
(A) Systemic-Behaviourist A software development company attempts to designlearning units about newtechnologies or softwareproducts for its developerstaff The design is drivenby a cycle of lsquocompetencygap analysisrsquo that modelsorganisational learning in terms of the aggregated competencies of the employeesmdashsimilar to theone described by Sicilia(2005)
Focuses on the individual where a body of informationforms the basis and rawmaterial for learning Learning is mainly considered as nonreflective ie based on skill learning and memorisation
(B) Sociocultural A special interest group ona concrete research topicaims at developing the knowledge of its membersby sharing experiencesand discussing on themesof common interest
Focuses on the social whereexperience is the main source of learning Conscious reflection on experience is transformed into learning
Semantic learning designs 347
copy 2006 The Author Journal compilation copy 2006 British Educational Communications and Technology Agency
of (B) In addition case (B) requires a model of social network that is largely irrelevantin case (A)
Differences in recording activity rationalesThe approach for the recording of design rationales also differs because of the differentcontexts Table 6 summarises the main elements for each of the scenarios
Conclusions and directions for researchCurrent technology and practice related to the description of reusable digital learningresources has resulted in techniques that allow the description of learning activities of
Table 5 Inputs activities and objectives of two learning design scenarios
Scenario Design inputs Activities and their objectives
(A) Systemic-Behaviourist A formal competency recordfor employees and competencyrequirements expressed in terms of amounts of required competencies (assuming a shared measurement scale) Such requirements are actually derived from strategy or from the contingent requirements of upcoming projects
Competency instances are used to measure the levelof Competency-Spec for the required competencies (or their constituents) A formal match considering the agenda of employees is carried out for determining the target learners LearningObjects forindividual use are selected asthe main elements of theactivities and the criteria forselection are again a match ofthe objectives stated inmetadata (as stated incategory 9 of Learning Object MetadatamdashClassification) and thecompetencies required
(B) Sociocultural The expertise of the special interest group members as stated in the concepts addressed by their publicationsprojects or other activities
Informal group activities are the main approach applicablesince there is not a predefinedgroup structure but a cultureof sharing and socialisation The results of the sharing of ideas can be expressed in terms of oc-knowsAboutpredicates and social ties will be modelled through positive valence FOAF-knows(httpxmlnscomfoaf01)predicates with an added numerical value representingstrength
348 British Journal of Educational Technology Vol 37 No 3 2006
copy 2006 The Author Journal compilation copy 2006 British Educational Communications and Technology Agency
a complex kind including their intended participants educational objectives and exe-cution constraints The representation and recording of the assumptions and guidelinesthat were considered in the creation of such resources would extend current techniquesto capture the rationale of the design itself This in turn will eventually result in theavailability of a large base of empirical data available in which theoretical assumptionsor rules are explicitly connected to the resulting designs as these could be linked withthe actual results and trace of the learning processes that used them The resulting bodyof evidence would represent a critical asset in the development of the theory of learningmediated through computers
This paper has described the first attempt to extend existing learning technology modelswith a schema explicitly targeted at describing learning design rationales The OpenCycknowledge base has been used as the semantic reference from which the ontologicalschema has been specified Concretely an activity-based schema based on IMS LD hasbeen described and theoretical assumptions and codified guidelines have been intro-duced The flexibility of the representation schema has been analysed through therepresentation of two stereotyped theoretical positions
Two main research directions should continue the work described here On the onehand further ontological engineering effort is required to come up with a more com-prehensive and rich schema for learning design rationales On the other hand tool
Table 6 Main elements of the two learning design scenarios
Scenario Approach Examples
(A) Systemic-Behaviourist Formal logical sentencesare used to guarantee that every design followsan organisationalobjective
It is required that the objective of every Method is attached to a Competency-Spec instance as objective
Formal assessment is a requirement foraccountability purposes
Guideline (GD1) is attached to everyDesigningLearningResource instance that produces an Act
(B) Sociocultural The interaction in informal settings can beused to derive social models (Sicilia amp Garciacutea2004)
The interactions in discussion canbe used to measure the degree of lsquoproximityrsquo between members Therationale of the activities can bejustified in terms of SocialTies inside SocialNetworks Further the focus of each groupactivity should be clearly stated anda combination of the increase in collective knowledge and the strengthening of social relations can be used for assessment
Semantic learning designs 349
copy 2006 The Author Journal compilation copy 2006 British Educational Communications and Technology Agency
implementation and record gathering efforts would be required to attain the ultimateobjective of the ontological definitions described so far With such effort an unprece-dented research material could be available in some years providing improved oppor-tunities for scientific inquiry on design as an activity regarding pedagogy
Other potential directions for research may be the description of interpretations of thedesign intentions (for which no rationale was explicitly recorded) the mining or detec-tion of patterns as related to rationale description or the detailed use of theoreticalrationales in agents that support the learning process which represent a category ofrelevant learning support software (Holmes in press)
ReferencesAllert H (2004) Coherent social systems for learningmdashan approach for contextualized and
community-centred metadata Journal of Interactive Media in Education 2004 1Baader F Calvanese D McGuinness D Nardi D amp Patel-Schneider P (Eds) (2003) The
description logic handbook Theory implementation and applications Cambridge CambridgeUniversity Press
Bannister D (1981) Personal construct theory and research method In P Reason amp J Rowan(Eds) Human inquiry a sourcebook of new paradigm research (pp 191ndash200) Chichester UKJohn Wiley and Sons
Berners-Lee T Hendler J amp Lassila O (2001) The semantic web Scientific American 284 534ndash43
Brown A amp Voltz B (2005) Elements of effective e-learning design International Review ofResearch in Open and Distance Learning 6 1
Buzza D Bean D Harrigan K amp Carey T (2005) Learning design repositories adaptinglearning design specifications for shared instructional knowledge Canadian Journal of Learningand Technology 30 3
Conole G Dyke M Oliver M amp Seale J (2004) Mapping pedagogy and tools for effectivelearning design Computers amp Education 43 1ndash2 17ndash33
Dodero J M Aedo I amp Diacuteaz P (2002) Participative knowledge production of learning objectsfor e-books The Electronic Library 20 4 296ndash305
Friesen N (2005) Interoperability and learning objects an overview of e-learning standardiza-tion Interdisciplinary Journal of Knowledge and Learning Objects 1 1 23ndash31
Gruber T (1995) Towards principles for the design of ontologies used for knowledge sharingInternational Journal of HumanndashComputer Studies 43 56 907ndash928
Hatchuel A (2002) Towards design theory and expandable rationality the unfinished programof Herbert Simon Journal of Management and Governance 5 3ndash4 260ndash271
Holmes J (in press) Designing agents to support learning by explaining Computers and EducationIEEE Learning Technology Standards Committee (2002) Learning object metadata (LOM) Final
Draft Standard IEEE 1484121ndash2002IMS (2003) IMS Learning Design Information Model (Version 10) Final Specification 20
January 2003Kaptelinin V Kuutti K amp Bannon L (1995) Activity theory basic concepts and applications
In Blumenthalet al (Eds) Humanndashcomputer interaction Lecture Notes in Computer Science 1015189ndash201
Koper R (2004) Use of the semantic web to solve some basic problems in education increaseflexible distributed lifelong learning decrease teacherrsquos workload Journal of Interactive Mediain Education 2004 6
Lenat D (1995) CYC a large-scale investment in knowledge infrastructure Communications ofthe ACM 38 11 33ndash38
350 British Journal of Educational Technology Vol 37 No 3 2006
copy 2006 The Author Journal compilation copy 2006 British Educational Communications and Technology Agency
McGreal R (2004) Learning objects a practical definition International Journal of InstructionalTechnology and Distance Learning 1 9 21ndash32
Packer M amp Goicoechea J (2000) Sociocultural and constructivist theories of learningontology not just epistemology Educational Psychologist 35 4 227ndash241
Palmer J (2001) Fifty major thinkers on education London and New York RoutledgePolsani P R (2003) Use and abuse of reusable learning objects Journal of Digital Information 3
4 Rothwell W amp Kazanas H (1992) Mastering the instructional design process San Francisco CA
Jossey-BassSaacutenchez-Alonso S amp Sicilia M A (2004) Relationships and commitments in learning object
metadata In Proceedings of the 5th International Conference on Information Technology BasedHigher Education and Training ITHET 2004 Istanbul Turkey
Sicilia M A (2005) Ontology-Based Competency Management Infrastructures for theKnowledge-intensive Learning Organization In Lytras amp Naeve (Eds) Intelligent Learning Infra-structures in Knowledge Intensive Organizations A Semantic Web perspective IDEA USA pp 302ndash324
Sicilia M A amp Garciacutea E (2004) Fuzzy group models for adaptation in cooperative informationretrieval contexts Lecture Notes in Computer Science 2932 324ndash334
Sicilia M A amp Garciacutea E (2005) On the convergence of formal ontologies and standardizede-learning Journal of Distance Education Technologies 3 2 13ndash29
Sicilia M A amp Lytras M (2005) On the representation of change according to different ontol-ogies of learning International Journal of Learning and Change 1 1 66ndash79
Sicilia M A Lytras M Rodriacuteguez E amp Garciacutea E (in press) Integrating descriptions of knowl-edge management learning activities into large ontological structures a case study Data andKnowledge Engineering
Tennant M (1991) Expertise as a dimension of adult development New Education 13 1 49ndash55Vanderdonckt J (1999) Development milestones towards a tool for working with guidelines
Interacting with Computers 12 2 81ndash118Wand Y amp Weber R (1995) On the deep structure of information systems Information Systems
Journal 5 203ndash223Wiley D A (2001) The instructional use of learning objects Bloomington IN Association for
Educational Communications and Technology
336
British Journal of Educational Technology Vol 37 No 3 2006
copy 2006 The Author Journal compilation copy 2006 British Educational Communications and Technology Agency
Table 1 Summary of the main specification definitions in the ontological schema
Concept Main descriptions Consequences in actual activities
Method-Spec Objectives and prerequisites(described in the followingsection) Structured in
play
specifications or directly
act
specifications
Instantiation is an independentmeaningful learning activity Playsare cotemporal
oc_subEvent
ssubsumed in the method Note thatthe semantic definition of subeventsyet entails the subsuming of subeventsin the containing event
Play-Spec Collection of sequential acts thatcomprise the play
Because plays are potentiallyconcurrent there are no furthertemporal restrictions
Act-Spec Collection of concurrent role parts that detail the activitiesincluded in the design Role partsand environments are the mainelements related to acts
The sequence of instantiations of actscould be sequenced through the
oc_startsAfterEndingOf
RolendashPart-Spec The linking of role specificationsto concrete activities or activitystructures
This entity is not required in theontological schema because theparticipation of roles in the activitiescan be stated in terms of specificationsof
oc_ActorRole
s or subeventsthat are not of a temporal nature
Activity-Spec The specification of learning andsupport structures in terms of
Role-Spec
s
ibtUsed LearningObject
s and used
Service
s Concrete objectivescan also be attached
Each instantiation of an
Activity-Spec
will result ininstances of specific types of
oc_Activity
Following LD specs
support
activities instantiate for everyrole mapping while
learning
onesinstantiate for all the role mappings
Environment These are descriptions of learning objects and services that do not require anexplicit mapping in the ontological schema
Service Service facilities have to belsquoinstantiated by a local run-timeservicesrsquo eg a chat or discussionforum Their computationalnature can be described interms of concrete
oc_ComputerProgramCW
Subcategories of such conceptualrepresentations of softwareentities can be used for thetypology of services
The
oc_ComputerProcessRunning
term is used to describe executionsof
oc_ComputerProgramCW
which cover every instance of aservice as understoodin IMS LD Thelinking between conceptual andactual running instances is coveredby the required predicate
oc-programOfProcess
Semantic learning designs
337
copy 2006 The Author Journal compilation copy 2006 British Educational Communications and Technology Agency
Learning object Those descriptions provided bylearning object metadataLearning objects can be definedas
oc_IBT
s as described inSicilia Lytras Rodriacuteguez andGarciacutea (in press) and approachesto describing metadata throughontologies yet existmdashsee an overview in Sicilia and Garciacutea(2005)
The learning objects selected arerepresented as instances in theontology which are declared to beused in activities through the
ibtUsed
predicate
Role-Spec Describe stereotypical actors oflearning and support activitiesThese act as lsquocontractrsquo for theactual participants in the senseprovided by Saacutenchez-Alonso andSicilia (2004)
Role specs are mapped to actualusers when the activities take placeThis is mapped to predicates of theclass
oc_ActorRole
Concept Main descriptions Consequences in actual activities
LD
=
Learning Design
Table 1 Continued
ontological perspective as the former allow onemdashand only onemdashinstance of the latterwith no particular logical differences
As described in Table 1 the
OpenCyc
built-in support for modelling temporal events canbe used to express the coordination relationships between the constituents of the unitsof learning The actual instances of learning events will be modelled as instances of
oc_Learning
(lsquothe collection of all events brief or extended in which an agent isacquiring information or know-howrsquo) and more specifically as instances of a speciali-sation
ComputerMediatedLearning
simply requiring the use of information net-works for at least part of the learning activities Because events in
OpenCyc
are alsosituations they have components in addition to the temporal extent including peopleinvolved information things used and other arrangements
Figure 1b depicts the main elements that describe actual current or past enactmentsof activities The schema is based on a clear separation of the specifications and theactual activities that by itself constitutes a metadata layering
OpenCyc
predicatesas
oc_situationConformsToSpecification
allows the linking of situations(every action is a situation also) to the specifications they conformmdashbe it purposely ornot The
LD-Activity term has been introduced as the common category of learningand support activities Methods plays and acts become special subcategories of LD-Activity and the constituent activities of Acts are instances of LD Activity also Rolesare attached to any kind of LD Activity even though it is required for the low-levelstructure The learning design concept of lsquoactivity structurersquo as an aggregation ofactivities does not require an explicit modelling because of the generic mereology ofactivities engineered in OpenCyc
338 British Journal of Educational Technology Vol 37 No 3 2006
copy 2006 The Author Journal compilation copy 2006 British Educational Communications and Technology Agency
Figure 1 (a) Summary of the competency components in the ontological schema (b) Summary of the main activityndashrole definitions in the ontological schema
oc_WorkingEvent
oc_IntelligentAgent
The competency elements aremeasurable characteristics andthey will be typed by the relatedSpecification classes eg Skill
SpecputlnPracticeln
(a)
Competency Competency-Spec
Competencyelement
Skill oc_ Competence
Attitude
Knowledge
has
thingSpecified
oc_Learning
ComputerMediatedLearning
oc_Role
(b)
Staff
Learner
oc_Specification
Method-Spec
Play-Spec
Act-Spec
Support
LD-Activity
contains
contains
Method
Play
Act
startAfterEndingOf
subAction
subAction
Activity-Spec
ltltActorSlotsgtgt
situationConformsToSpec
Activity-Spec
Activity
Semantic learning designs 339
copy 2006 The Author Journal compilation copy 2006 British Educational Communications and Technology Agency
The concept of oc_Role accounts not only for participant roles but also for otherkinds of lsquocomponentsrsquo This is why two specialisations of the concept ofoc_ActorSlot labelled Staff and Learner are required to differentiate partici-pants from other elements As is obvious from the earlier discussion the built-in rep-resentation of activities in OpenCyc is powerful enough for the definition of learningdesign activities
The final part of the ontological specification described so far is that of the mappingfrom LD specifications to the actual past or present activity realisations Theoc_situationConformsToSpecification predicate can be used to trace whichactivity realizations come from a given learning design activity definition and the sameoccurs with the oc_thingSpecified for other learning design elements such asroles or environments Existing learning design execution engines as CopperCore(httpcoppercoreorg) actually map LD specifications to online collaboration so thatthe translation of the ontological representations of learning designs to ontology defi-nitions of activities is straightforward following similar representation
Describing objectives outcomes and preconditionsAccording to IMS learning design learning objectives and prerequisites lsquocan bedescribed using the IMS Reusable Definition of Competency or Educational Objective(RDCEO) format but can also refer to simple resources (eg a text) with a descriptionof the learning objectiversquo As the latter option is not useful from the viewpoint of com-putational semantics we will adhere to the former As commented earlier the compe-tency view is not exclusive to other kinds of objectives related to other ontologicalstandpoints (Sicilia amp Lytras 2005) and other conceptions of change or improvementmay include social ties cognitive structures or the combination of the knowledge of agroup In fact competencies adhere mostly to behaviouristic frameworks that focus onobservable external behaviour (Tennant 1991) rather than on cognitive representa-tions Nonetheless the further discussion is equally applicable to other kinds of objec-tives so that we will adhere to the competency view
oc_Competence in OpenCyc is defined as a quantity that lsquofor describing the level ofskill with which an agent performs some taskrsquo This does not remove the lack of preci-sion that is yet present in RDCEO definitions as described by Sicilia (2005) As analternative Figure 1a provides a simplified account of the model described by Sicilia asrelated to OpenCyc terms in which competence is integrated as one of the elements of thebroader notion of competency according to the model of Rummel (Rothwell amp Kazanas1992) which considers work situations attributes of the individual that are applied insuch situations and measurement with respect to some standard of performance Themost relevant distinction is again the separation of actual competency representationsand competency lsquodefinitionsrsquo (ie specifications) intended to provide a type structure tocompetencies to be used in objective or precondition specification In this case thegeneric oc-thingSpecified predicate can be used for the linking of both levels ofrepresentation
340 British Journal of Educational Technology Vol 37 No 3 2006
copy 2006 The Author Journal compilation copy 2006 British Educational Communications and Technology Agency
Objectives and preconditions can then be expressed as formulas on the acquisition ofcompetencies or their components (competency elements) Descriptions specific toOpenCyc may use oc_CycLSentence-Assertible but simpler enumeration ofcompetency components could be used instead
Describing learning design rationales theories hypotheses guidelines and rulesThe concept of lsquodesignrsquo for our purposes is properly captured in the 5a definitionprovided in the Merriam Webster Online Dictionary (wwwm-wcom) lsquoan underlyingscheme that governs functioning developing or unfoldingrsquo In consequence learningdesigns refer to artefacts and not to the process of devising the educational solutionThis is equivalent to the concept oc_Design subsumed by the notion ofoc_Specification which is described as lsquoan abstract work that constitutes adescription of the properties of a Situation or a SomethingExisting and sometimeseven entire collections of such thingsrsquo In consequence the term LearningDesignwould simply delineate a subset of specifications that are oriented to describe howlearning experiences with concrete objectives should be Trivially oc_Designerdescribes the authors of the learning designs and oc_Designing represent theoc_HumanActivity that leads to the production of learning designs Designing as aspecial kind of activity is carried out (oc_doneBy) by a designer which provides a wayto attach metainformation about the rationale of the design
From an ontological perspective the departing problem in modelling learning designrationales is that their assumptions and hypotheses of different theoretical standpointsneed not be compatible or need not consider the same aspects of reality (Conole et al2004 Packer amp Goicoechea 2000) Furthermore different pedagogical approaches areknown to result in different mappings to activities (Conole et al 2004) This wouldeither require the provision of separate ontologies or the use of a representationalmechanism that allows such kind of potential inconsistency or divergence The conceptof microtheory in OpenCyc provides such representational mechanism intended to orga-nise assertions that depend on lsquoshared set of assumptions on which the truth of theassertions dependsrsquo Definitions inside the same microtheory need to be consistent butthis is not required across microtheories The descriptions provided in the earlier sectioncan be used as a base microtheory called LearningActivitiesMt for others that are spe-cific to concrete theoretical frameworks In this section we will describe elements thatare common to the description of theoretical frameworks the distinction among micro-theories will not be discussed
The intellectual processes of learning design in themselves will be represented asoc_Designing activities carried out by oc_Designers An additional consider-ation is that the information about the rationale of the design would be better attachedto representations of each of the design processes (ie to oc_Designing instances)than to the final design This enables the registering of different assumptions for differ-ent contributions to the same artefact eg different revisions from the same of differentdesigners
Semantic learning designs 341
copy 2006 The Author Journal compilation copy 2006 British Educational Communications and Technology Agency
Several designers may participate in the same designing activity and formal frameworksfor collaborative learning resource creation (Dodero Aedo amp Diacuteaz 2002) can be inte-grated by clearly specifying which class of things can be the outcome of a designingprocess For the sake of flexibility the concept of LearningDesignPart is introduced as anall-encompassing category that includes each of the elements described in the previoussection This includes not only activities and roles but also objectives conditions andthe like Furthermore the granularity of the descriptions allows for different levels fromconsidering a unique designing activity for the whole package to recording each ele-ment of the learning design comprehensively Because oc_Actions are subsumed bythe concept of oc_Event considered as a lsquodynamic situation in which the state of theworld changesrsquo each edition of a metadata record or any other form of description maybe subject to be described The concept is flexible enough to be able to record fine-grained detailed decisions This of course may include different versions of the sameartefacts but this is an orthogonal aspect of configuration that is not relevant to ourpresent discussion
Figure 2 depicts the main elements of the loci to which rationales assumptions andguidelines will be attached
A LearningDesigner can be defined as the class of humans that performactivities of learning resource design The consideration of learning design as aunique human activity could be considered as restrictive of intelligent softwarefor the task Simply changing the oc_Designer class to a subconcept ofoc_IntelligentAgent could remove this but this entails a change in Open-Cyc knowledge base The concept oc_Designing is a subclass of the conceptoc_ThinkingAboutTheStateOfTheWorld which represents lsquothinking pro-cessesrsquo which include mental events such as planning evaluating or inventingThus the concept of design described so far should be understood in terms of the
Figure 2 Main elements of the ontological schema for the loci of design rationales
oc_performedBy modifies
342 British Journal of Educational Technology Vol 37 No 3 2006
copy 2006 The Author Journal compilation copy 2006 British Educational Communications and Technology Agency
concept of lsquoexpanded rationalityrsquo as described by Hatchuel (2002) integrating cre-ativity and unexpected expansions of the original requirements This precludes onto-logical definitions in which the problem space is completely bounded a priori Inconsequence a degree of openness is necessary to integrate different kinds of detailin description from fully described ones to others with shallower semantics eg someproviding only references to generic assumptions
Theoretical frameworksWe will make the distinction that guidelines (as further defined) are the propositions thatdirectly predicate about the validity or appropriateness of learning design parts whileother theoretical propositions about learning that are not directly saying anythingabout the design of digital learning resources are considered as assumptions followingthe general sense provided in the Merriam-Webster dictionary lsquoa fact or statement (asa proposition axiom postulate or notion) taken for grantedrsquo
The fact that some assumption hypotheses or proposed law about learning is sustainedby a broader empirical base or is better sustained by rational means is out of the scopeof this work because here we focus on design practice (even though it is of an obviousinterest for researchers) In principle even unorthodox positions may be sustainedbecause the mechanism of microtheories and their inclusion mechanism allows forgreat flexibility in modelling them
Table 2 provides some examples of theoretical assumptions and related guidelines forlsquoextremersquo archetypical standpoints on learning These are not intended to reflect actualcommon positions but they are provided as illustrative examples of radical positionsthat could be modelled with our ontological schema
The examples provided in Table 2 go a step further than toolkits in Conole et al (2004)in making the design rationale explicit Nonetheless in many cases lsquocommon sensersquo isapplied to design and a shallower approach to specification could be used instead egproviding simply links to the theoretical standpoints used It should be noted also thatthe resulting guidelines vary greatly in their level of concreteness and some of themmay be considered as mandatory from some standpoints while others are mere generalrepresentations This would lead to different computational strategies
Guidelines and rulesThe WordNet (httpwordnetprincetonedu) thesaurus defines guidelines as lsquoguidancerelative to setting standards or determining a course of actionrsquo or lsquoa rule or principlethat provides guidance to appropriate behaviorrsquo In our case guidelines are statementsthat serve as criteria for decision in designing learning Given that e-learning designis still far from being a mature discipline guidelinesmdashas instances of recordedknowledgemdashcan be characterised as oc_NormativeSpecifications becausethey describe how learning resource design activities and their outcomes lsquoshould bersquoin a propositional form Propositions lsquohave some truth value in some context or
Semantic learning designs 343
copy 2006 The Author Journal compilation copy 2006 British Educational Communications and Technology Agency
ldquoworldrdquo rsquo Furthermore propositions lsquoare assumed to be representable (at least in prin-ciple) by a sentence in some formal or natural languagersquo This raises the need for alanguage that expresses guidelines in a general sense There are two main options forsuch representation be them formal logical sentences or reified guidelines
Formal logical sentence representation entails the formalisation through sentences ofthe constraints of the designs For example if lsquoevery learning resource must be activersquoaccording to some theoretical standpoint (eg one that fosters reflective learning) thissentence may be represented as an axiom and put into the appropriate separate micro-theory as the following CycL fragment
($forAll X ($implies
($isa X $LearningObject)
($resourceType X $Active))
Then resources that are not active are simply excluded from the notion itself This is astrong position in the sense that precludes some views on the learning design process
Table 2 Some examples of theoretical assumptions and guidelines
Archetypicalstandpoint Assumptions Guidelines
Instructivist(Behaviouristic)
bull Learning occurs through association and reinforcement
bull The results of learning can be appreciated through observable outcomes
(GB1) Every instance of oc-Learning entails theexposure to some LearningObjectthat produces the association (GB2) Learning Activitysequences should provide assessmentactivities that confirm theassociations that should have beenestablished so far
Constructivist bull Learning proceeds by building mental structures
(GC1) Agentrsquos mental structuresshould be represented(GC2) These structures should bepre-assessed in order to becomparable with the resulting ones(GC3) Design Activities arerecommended(GC4) Contrast Activities arerecommended
Sociocultural bull Learning occurs through social participatio
(GSC1) Every instance ofoc-Learning entails theinteraction of severaloc-IntelligentAgents asoc-Actors
344 British Journal of Educational Technology Vol 37 No 3 2006
copy 2006 The Author Journal compilation copy 2006 British Educational Communications and Technology Agency
Another typical example may be that of a sociocultural guideline that preclude Methodsthat do not entail multirole activities that provide access to social interaction ServicesThis approach to guidelines provides the best framework for automation and wouldenable the creation of microtheories that represent clearly formulated views onlearning Nonetheless there are two major problems for the use of this option as theunique mechanism for encoding design rationales
1 There is no clear-cut distinction between models and theories of learning becausethey come from a history of schools of thought with interrelations (Palmer 2001)This would lead to the need for the representation of dozens of variants reflectingpositions that only diverge in some aspects
2 Many guidelines cannot be represented as logical sentences because theyrepresent lsquorecommendationsrsquo or vague requirements eg see (GC3) or (GC4)(Table 2)
These issues suggest that a combination of formal sentences with reified guidelinesprovide the best option for flexibility The representation of guidelines has beenstudied in the field of HumanndashComputer-Interaction (HCI) in the last decade(Vanderdonckt 1999) and the main philosophy of such representations remainsvalid in the learning design domain The major challenge for such representations isthat of providing computational semantics that could be used for the implementationof software support but this is difficult to attain in all the cases Thus the concept oflearning design guideline will provide two kinds of representations general andschema-specific
The general use of instances of oc_NormativeSpecification allows the model-ling of lsquoconceptual works that describe how something should bersquo This may includerepresentation of known normative representations or procedures For example onemay refer to a concrete learning design model as that derived from Bannister (1981)which states that creativity cycles follow an initial phase of brainstorming (idea cap-ture) followed by a phase in which key issues are focused into an operation strategy(idea development) and a third phase in which project control takes place through alsquorecursive cyclersquo (operational management) The reference could be made by pointing tothe academic reference describing it as a conceptual work or it could be described asan Activity-Spec modelled explicitly inside the ontology As another examplesome detailed guidelines could be modelled by the normative specification sub-category of oc-Instructions defined as lsquoa sequence of tasks to be performed byan agentrsquo
Schema-specific guidelines restrict the freedom to represent criteria as conceptual workinstances to a concrete and commonly agreed ontology that determine what can be saidabout how learning designs should be according to a theoretical standpoint or frame-work These build on an activityndashrole ontology as the one described in the previoussection and add a prescription on the kind of lsquoshould-bersquo predications that are allow-able Table 3 describes examples of a minimal language that could be used for thatpurpose
Semantic learning designs 345
copy 2006 The Author Journal compilation copy 2006 British Educational Communications and Technology Agency
In the examples CAT represents a category of individual expressed through a conceptor a logical formula and the sentence is a synthetic way of expressing some recom-mended or mandatory constraint over such category Other languages could be engi-neered in further research but the one described can be directly expressed in OpenCycand combined with software modules that check or enforce them in some way Notethat this representation does not use direct OpenCyc sentences so that reification is usedinstead of direct logical sentences It is not the aim of this paper to determine the mostpowerful and flexible representation language but only set a foundation for furtherspecification efforts
Analysis and comparison of two theoretical semantic frameworksAs an assessment of the representational capabilities of the scheme just described acase study of representation is sketched in this section We will use the learning designtoolkit described by Conole et al (2004) as the underlying terminology even thoughother frameworks could be used instead
Two rather diverging positions for design will be used as an illustration Table 4 sum-marises the two design scenarios
Table 3 Possible language for use in learning designs
Sentence type Potential uses Example
RecommendationshouldBe (CATSENT)
Recommend kinds ofactivities
Design activities are recommended (GC3)shouldBe(Method subEvent (CATDesignActivity))
ObligationmustBe (CATSENT)
Some activity typesmust follow others
Assessments must follow each expositionto contents (GD1) mustBe(and(ExpositiveActivityIndividualActivity)startAfterEndingOf(CAT
AssessmentActivity)
Activities must havesome characteristic
Every activity must entail interactionmustBe(LDActivity(usedIn(CAT
InteractiveService)))Every activity must provide acontent-oriented learning object forindividual use (similar to GB1)mustBe(LDActivity(and (usedIn(CAT
LearningObject)(learnerRoleNumber(1))))
Some structure shouldbe enforced in methods
Methods should have a sequential actstructuremustBe(Methodnot ((subEvent(CAT Play))))
CAT = category GC3 = Constructivist Guideline 3 (see Table 2) GD1 = Guideline 1
346 British Journal of Educational Technology Vol 37 No 3 2006
copy 2006 The Author Journal compilation copy 2006 British Educational Communications and Technology Agency
The following subsections sketch the main outcomes of an example learning designprocess for both scenarios assuming that the process described by Conole et al (2004)is used in both situations
Differences in modelling the design processThe forces driving the design process in both scenarios are fairly diverging The designof Scenario A will typically be constrained by a clear target specified in terms of anaggregated competency level while in Scenario B there are no concrete competenciesto be acquired and the main objective of the design activity would be twofold On theone hand the activities are targeted at promoting interchange of ideas in an attemptto stimulate casual convergence of interest which is the main value of the culture ofthe special interest group On the other hand the activities have the generic objectiveof strengthening social ties which is an important element of cohesion for the under-taking of joint projects
Table 5 describes the main inputs and possible activities and their objectives at a highlevel
The changes intended to be achieved in each of the scenarios are stated differently witha formal definition in (A) and a shallow cognitive and social representation in the case
Table 4 Learning design scenarios
Scenario DescriptionClassification according to
Conole et al (2004) toolkit
(A) Systemic-Behaviourist A software development company attempts to designlearning units about newtechnologies or softwareproducts for its developerstaff The design is drivenby a cycle of lsquocompetencygap analysisrsquo that modelsorganisational learning in terms of the aggregated competencies of the employeesmdashsimilar to theone described by Sicilia(2005)
Focuses on the individual where a body of informationforms the basis and rawmaterial for learning Learning is mainly considered as nonreflective ie based on skill learning and memorisation
(B) Sociocultural A special interest group ona concrete research topicaims at developing the knowledge of its membersby sharing experiencesand discussing on themesof common interest
Focuses on the social whereexperience is the main source of learning Conscious reflection on experience is transformed into learning
Semantic learning designs 347
copy 2006 The Author Journal compilation copy 2006 British Educational Communications and Technology Agency
of (B) In addition case (B) requires a model of social network that is largely irrelevantin case (A)
Differences in recording activity rationalesThe approach for the recording of design rationales also differs because of the differentcontexts Table 6 summarises the main elements for each of the scenarios
Conclusions and directions for researchCurrent technology and practice related to the description of reusable digital learningresources has resulted in techniques that allow the description of learning activities of
Table 5 Inputs activities and objectives of two learning design scenarios
Scenario Design inputs Activities and their objectives
(A) Systemic-Behaviourist A formal competency recordfor employees and competencyrequirements expressed in terms of amounts of required competencies (assuming a shared measurement scale) Such requirements are actually derived from strategy or from the contingent requirements of upcoming projects
Competency instances are used to measure the levelof Competency-Spec for the required competencies (or their constituents) A formal match considering the agenda of employees is carried out for determining the target learners LearningObjects forindividual use are selected asthe main elements of theactivities and the criteria forselection are again a match ofthe objectives stated inmetadata (as stated incategory 9 of Learning Object MetadatamdashClassification) and thecompetencies required
(B) Sociocultural The expertise of the special interest group members as stated in the concepts addressed by their publicationsprojects or other activities
Informal group activities are the main approach applicablesince there is not a predefinedgroup structure but a cultureof sharing and socialisation The results of the sharing of ideas can be expressed in terms of oc-knowsAboutpredicates and social ties will be modelled through positive valence FOAF-knows(httpxmlnscomfoaf01)predicates with an added numerical value representingstrength
348 British Journal of Educational Technology Vol 37 No 3 2006
copy 2006 The Author Journal compilation copy 2006 British Educational Communications and Technology Agency
a complex kind including their intended participants educational objectives and exe-cution constraints The representation and recording of the assumptions and guidelinesthat were considered in the creation of such resources would extend current techniquesto capture the rationale of the design itself This in turn will eventually result in theavailability of a large base of empirical data available in which theoretical assumptionsor rules are explicitly connected to the resulting designs as these could be linked withthe actual results and trace of the learning processes that used them The resulting bodyof evidence would represent a critical asset in the development of the theory of learningmediated through computers
This paper has described the first attempt to extend existing learning technology modelswith a schema explicitly targeted at describing learning design rationales The OpenCycknowledge base has been used as the semantic reference from which the ontologicalschema has been specified Concretely an activity-based schema based on IMS LD hasbeen described and theoretical assumptions and codified guidelines have been intro-duced The flexibility of the representation schema has been analysed through therepresentation of two stereotyped theoretical positions
Two main research directions should continue the work described here On the onehand further ontological engineering effort is required to come up with a more com-prehensive and rich schema for learning design rationales On the other hand tool
Table 6 Main elements of the two learning design scenarios
Scenario Approach Examples
(A) Systemic-Behaviourist Formal logical sentencesare used to guarantee that every design followsan organisationalobjective
It is required that the objective of every Method is attached to a Competency-Spec instance as objective
Formal assessment is a requirement foraccountability purposes
Guideline (GD1) is attached to everyDesigningLearningResource instance that produces an Act
(B) Sociocultural The interaction in informal settings can beused to derive social models (Sicilia amp Garciacutea2004)
The interactions in discussion canbe used to measure the degree of lsquoproximityrsquo between members Therationale of the activities can bejustified in terms of SocialTies inside SocialNetworks Further the focus of each groupactivity should be clearly stated anda combination of the increase in collective knowledge and the strengthening of social relations can be used for assessment
Semantic learning designs 349
copy 2006 The Author Journal compilation copy 2006 British Educational Communications and Technology Agency
implementation and record gathering efforts would be required to attain the ultimateobjective of the ontological definitions described so far With such effort an unprece-dented research material could be available in some years providing improved oppor-tunities for scientific inquiry on design as an activity regarding pedagogy
Other potential directions for research may be the description of interpretations of thedesign intentions (for which no rationale was explicitly recorded) the mining or detec-tion of patterns as related to rationale description or the detailed use of theoreticalrationales in agents that support the learning process which represent a category ofrelevant learning support software (Holmes in press)
ReferencesAllert H (2004) Coherent social systems for learningmdashan approach for contextualized and
community-centred metadata Journal of Interactive Media in Education 2004 1Baader F Calvanese D McGuinness D Nardi D amp Patel-Schneider P (Eds) (2003) The
description logic handbook Theory implementation and applications Cambridge CambridgeUniversity Press
Bannister D (1981) Personal construct theory and research method In P Reason amp J Rowan(Eds) Human inquiry a sourcebook of new paradigm research (pp 191ndash200) Chichester UKJohn Wiley and Sons
Berners-Lee T Hendler J amp Lassila O (2001) The semantic web Scientific American 284 534ndash43
Brown A amp Voltz B (2005) Elements of effective e-learning design International Review ofResearch in Open and Distance Learning 6 1
Buzza D Bean D Harrigan K amp Carey T (2005) Learning design repositories adaptinglearning design specifications for shared instructional knowledge Canadian Journal of Learningand Technology 30 3
Conole G Dyke M Oliver M amp Seale J (2004) Mapping pedagogy and tools for effectivelearning design Computers amp Education 43 1ndash2 17ndash33
Dodero J M Aedo I amp Diacuteaz P (2002) Participative knowledge production of learning objectsfor e-books The Electronic Library 20 4 296ndash305
Friesen N (2005) Interoperability and learning objects an overview of e-learning standardiza-tion Interdisciplinary Journal of Knowledge and Learning Objects 1 1 23ndash31
Gruber T (1995) Towards principles for the design of ontologies used for knowledge sharingInternational Journal of HumanndashComputer Studies 43 56 907ndash928
Hatchuel A (2002) Towards design theory and expandable rationality the unfinished programof Herbert Simon Journal of Management and Governance 5 3ndash4 260ndash271
Holmes J (in press) Designing agents to support learning by explaining Computers and EducationIEEE Learning Technology Standards Committee (2002) Learning object metadata (LOM) Final
Draft Standard IEEE 1484121ndash2002IMS (2003) IMS Learning Design Information Model (Version 10) Final Specification 20
January 2003Kaptelinin V Kuutti K amp Bannon L (1995) Activity theory basic concepts and applications
In Blumenthalet al (Eds) Humanndashcomputer interaction Lecture Notes in Computer Science 1015189ndash201
Koper R (2004) Use of the semantic web to solve some basic problems in education increaseflexible distributed lifelong learning decrease teacherrsquos workload Journal of Interactive Mediain Education 2004 6
Lenat D (1995) CYC a large-scale investment in knowledge infrastructure Communications ofthe ACM 38 11 33ndash38
350 British Journal of Educational Technology Vol 37 No 3 2006
copy 2006 The Author Journal compilation copy 2006 British Educational Communications and Technology Agency
McGreal R (2004) Learning objects a practical definition International Journal of InstructionalTechnology and Distance Learning 1 9 21ndash32
Packer M amp Goicoechea J (2000) Sociocultural and constructivist theories of learningontology not just epistemology Educational Psychologist 35 4 227ndash241
Palmer J (2001) Fifty major thinkers on education London and New York RoutledgePolsani P R (2003) Use and abuse of reusable learning objects Journal of Digital Information 3
4 Rothwell W amp Kazanas H (1992) Mastering the instructional design process San Francisco CA
Jossey-BassSaacutenchez-Alonso S amp Sicilia M A (2004) Relationships and commitments in learning object
metadata In Proceedings of the 5th International Conference on Information Technology BasedHigher Education and Training ITHET 2004 Istanbul Turkey
Sicilia M A (2005) Ontology-Based Competency Management Infrastructures for theKnowledge-intensive Learning Organization In Lytras amp Naeve (Eds) Intelligent Learning Infra-structures in Knowledge Intensive Organizations A Semantic Web perspective IDEA USA pp 302ndash324
Sicilia M A amp Garciacutea E (2004) Fuzzy group models for adaptation in cooperative informationretrieval contexts Lecture Notes in Computer Science 2932 324ndash334
Sicilia M A amp Garciacutea E (2005) On the convergence of formal ontologies and standardizede-learning Journal of Distance Education Technologies 3 2 13ndash29
Sicilia M A amp Lytras M (2005) On the representation of change according to different ontol-ogies of learning International Journal of Learning and Change 1 1 66ndash79
Sicilia M A Lytras M Rodriacuteguez E amp Garciacutea E (in press) Integrating descriptions of knowl-edge management learning activities into large ontological structures a case study Data andKnowledge Engineering
Tennant M (1991) Expertise as a dimension of adult development New Education 13 1 49ndash55Vanderdonckt J (1999) Development milestones towards a tool for working with guidelines
Interacting with Computers 12 2 81ndash118Wand Y amp Weber R (1995) On the deep structure of information systems Information Systems
Journal 5 203ndash223Wiley D A (2001) The instructional use of learning objects Bloomington IN Association for
Educational Communications and Technology
Semantic learning designs
337
copy 2006 The Author Journal compilation copy 2006 British Educational Communications and Technology Agency
Learning object Those descriptions provided bylearning object metadataLearning objects can be definedas
oc_IBT
s as described inSicilia Lytras Rodriacuteguez andGarciacutea (in press) and approachesto describing metadata throughontologies yet existmdashsee an overview in Sicilia and Garciacutea(2005)
The learning objects selected arerepresented as instances in theontology which are declared to beused in activities through the
ibtUsed
predicate
Role-Spec Describe stereotypical actors oflearning and support activitiesThese act as lsquocontractrsquo for theactual participants in the senseprovided by Saacutenchez-Alonso andSicilia (2004)
Role specs are mapped to actualusers when the activities take placeThis is mapped to predicates of theclass
oc_ActorRole
Concept Main descriptions Consequences in actual activities
LD
=
Learning Design
Table 1 Continued
ontological perspective as the former allow onemdashand only onemdashinstance of the latterwith no particular logical differences
As described in Table 1 the
OpenCyc
built-in support for modelling temporal events canbe used to express the coordination relationships between the constituents of the unitsof learning The actual instances of learning events will be modelled as instances of
oc_Learning
(lsquothe collection of all events brief or extended in which an agent isacquiring information or know-howrsquo) and more specifically as instances of a speciali-sation
ComputerMediatedLearning
simply requiring the use of information net-works for at least part of the learning activities Because events in
OpenCyc
are alsosituations they have components in addition to the temporal extent including peopleinvolved information things used and other arrangements
Figure 1b depicts the main elements that describe actual current or past enactmentsof activities The schema is based on a clear separation of the specifications and theactual activities that by itself constitutes a metadata layering
OpenCyc
predicatesas
oc_situationConformsToSpecification
allows the linking of situations(every action is a situation also) to the specifications they conformmdashbe it purposely ornot The
LD-Activity term has been introduced as the common category of learningand support activities Methods plays and acts become special subcategories of LD-Activity and the constituent activities of Acts are instances of LD Activity also Rolesare attached to any kind of LD Activity even though it is required for the low-levelstructure The learning design concept of lsquoactivity structurersquo as an aggregation ofactivities does not require an explicit modelling because of the generic mereology ofactivities engineered in OpenCyc
338 British Journal of Educational Technology Vol 37 No 3 2006
copy 2006 The Author Journal compilation copy 2006 British Educational Communications and Technology Agency
Figure 1 (a) Summary of the competency components in the ontological schema (b) Summary of the main activityndashrole definitions in the ontological schema
oc_WorkingEvent
oc_IntelligentAgent
The competency elements aremeasurable characteristics andthey will be typed by the relatedSpecification classes eg Skill
SpecputlnPracticeln
(a)
Competency Competency-Spec
Competencyelement
Skill oc_ Competence
Attitude
Knowledge
has
thingSpecified
oc_Learning
ComputerMediatedLearning
oc_Role
(b)
Staff
Learner
oc_Specification
Method-Spec
Play-Spec
Act-Spec
Support
LD-Activity
contains
contains
Method
Play
Act
startAfterEndingOf
subAction
subAction
Activity-Spec
ltltActorSlotsgtgt
situationConformsToSpec
Activity-Spec
Activity
Semantic learning designs 339
copy 2006 The Author Journal compilation copy 2006 British Educational Communications and Technology Agency
The concept of oc_Role accounts not only for participant roles but also for otherkinds of lsquocomponentsrsquo This is why two specialisations of the concept ofoc_ActorSlot labelled Staff and Learner are required to differentiate partici-pants from other elements As is obvious from the earlier discussion the built-in rep-resentation of activities in OpenCyc is powerful enough for the definition of learningdesign activities
The final part of the ontological specification described so far is that of the mappingfrom LD specifications to the actual past or present activity realisations Theoc_situationConformsToSpecification predicate can be used to trace whichactivity realizations come from a given learning design activity definition and the sameoccurs with the oc_thingSpecified for other learning design elements such asroles or environments Existing learning design execution engines as CopperCore(httpcoppercoreorg) actually map LD specifications to online collaboration so thatthe translation of the ontological representations of learning designs to ontology defi-nitions of activities is straightforward following similar representation
Describing objectives outcomes and preconditionsAccording to IMS learning design learning objectives and prerequisites lsquocan bedescribed using the IMS Reusable Definition of Competency or Educational Objective(RDCEO) format but can also refer to simple resources (eg a text) with a descriptionof the learning objectiversquo As the latter option is not useful from the viewpoint of com-putational semantics we will adhere to the former As commented earlier the compe-tency view is not exclusive to other kinds of objectives related to other ontologicalstandpoints (Sicilia amp Lytras 2005) and other conceptions of change or improvementmay include social ties cognitive structures or the combination of the knowledge of agroup In fact competencies adhere mostly to behaviouristic frameworks that focus onobservable external behaviour (Tennant 1991) rather than on cognitive representa-tions Nonetheless the further discussion is equally applicable to other kinds of objec-tives so that we will adhere to the competency view
oc_Competence in OpenCyc is defined as a quantity that lsquofor describing the level ofskill with which an agent performs some taskrsquo This does not remove the lack of preci-sion that is yet present in RDCEO definitions as described by Sicilia (2005) As analternative Figure 1a provides a simplified account of the model described by Sicilia asrelated to OpenCyc terms in which competence is integrated as one of the elements of thebroader notion of competency according to the model of Rummel (Rothwell amp Kazanas1992) which considers work situations attributes of the individual that are applied insuch situations and measurement with respect to some standard of performance Themost relevant distinction is again the separation of actual competency representationsand competency lsquodefinitionsrsquo (ie specifications) intended to provide a type structure tocompetencies to be used in objective or precondition specification In this case thegeneric oc-thingSpecified predicate can be used for the linking of both levels ofrepresentation
340 British Journal of Educational Technology Vol 37 No 3 2006
copy 2006 The Author Journal compilation copy 2006 British Educational Communications and Technology Agency
Objectives and preconditions can then be expressed as formulas on the acquisition ofcompetencies or their components (competency elements) Descriptions specific toOpenCyc may use oc_CycLSentence-Assertible but simpler enumeration ofcompetency components could be used instead
Describing learning design rationales theories hypotheses guidelines and rulesThe concept of lsquodesignrsquo for our purposes is properly captured in the 5a definitionprovided in the Merriam Webster Online Dictionary (wwwm-wcom) lsquoan underlyingscheme that governs functioning developing or unfoldingrsquo In consequence learningdesigns refer to artefacts and not to the process of devising the educational solutionThis is equivalent to the concept oc_Design subsumed by the notion ofoc_Specification which is described as lsquoan abstract work that constitutes adescription of the properties of a Situation or a SomethingExisting and sometimeseven entire collections of such thingsrsquo In consequence the term LearningDesignwould simply delineate a subset of specifications that are oriented to describe howlearning experiences with concrete objectives should be Trivially oc_Designerdescribes the authors of the learning designs and oc_Designing represent theoc_HumanActivity that leads to the production of learning designs Designing as aspecial kind of activity is carried out (oc_doneBy) by a designer which provides a wayto attach metainformation about the rationale of the design
From an ontological perspective the departing problem in modelling learning designrationales is that their assumptions and hypotheses of different theoretical standpointsneed not be compatible or need not consider the same aspects of reality (Conole et al2004 Packer amp Goicoechea 2000) Furthermore different pedagogical approaches areknown to result in different mappings to activities (Conole et al 2004) This wouldeither require the provision of separate ontologies or the use of a representationalmechanism that allows such kind of potential inconsistency or divergence The conceptof microtheory in OpenCyc provides such representational mechanism intended to orga-nise assertions that depend on lsquoshared set of assumptions on which the truth of theassertions dependsrsquo Definitions inside the same microtheory need to be consistent butthis is not required across microtheories The descriptions provided in the earlier sectioncan be used as a base microtheory called LearningActivitiesMt for others that are spe-cific to concrete theoretical frameworks In this section we will describe elements thatare common to the description of theoretical frameworks the distinction among micro-theories will not be discussed
The intellectual processes of learning design in themselves will be represented asoc_Designing activities carried out by oc_Designers An additional consider-ation is that the information about the rationale of the design would be better attachedto representations of each of the design processes (ie to oc_Designing instances)than to the final design This enables the registering of different assumptions for differ-ent contributions to the same artefact eg different revisions from the same of differentdesigners
Semantic learning designs 341
copy 2006 The Author Journal compilation copy 2006 British Educational Communications and Technology Agency
Several designers may participate in the same designing activity and formal frameworksfor collaborative learning resource creation (Dodero Aedo amp Diacuteaz 2002) can be inte-grated by clearly specifying which class of things can be the outcome of a designingprocess For the sake of flexibility the concept of LearningDesignPart is introduced as anall-encompassing category that includes each of the elements described in the previoussection This includes not only activities and roles but also objectives conditions andthe like Furthermore the granularity of the descriptions allows for different levels fromconsidering a unique designing activity for the whole package to recording each ele-ment of the learning design comprehensively Because oc_Actions are subsumed bythe concept of oc_Event considered as a lsquodynamic situation in which the state of theworld changesrsquo each edition of a metadata record or any other form of description maybe subject to be described The concept is flexible enough to be able to record fine-grained detailed decisions This of course may include different versions of the sameartefacts but this is an orthogonal aspect of configuration that is not relevant to ourpresent discussion
Figure 2 depicts the main elements of the loci to which rationales assumptions andguidelines will be attached
A LearningDesigner can be defined as the class of humans that performactivities of learning resource design The consideration of learning design as aunique human activity could be considered as restrictive of intelligent softwarefor the task Simply changing the oc_Designer class to a subconcept ofoc_IntelligentAgent could remove this but this entails a change in Open-Cyc knowledge base The concept oc_Designing is a subclass of the conceptoc_ThinkingAboutTheStateOfTheWorld which represents lsquothinking pro-cessesrsquo which include mental events such as planning evaluating or inventingThus the concept of design described so far should be understood in terms of the
Figure 2 Main elements of the ontological schema for the loci of design rationales
oc_performedBy modifies
342 British Journal of Educational Technology Vol 37 No 3 2006
copy 2006 The Author Journal compilation copy 2006 British Educational Communications and Technology Agency
concept of lsquoexpanded rationalityrsquo as described by Hatchuel (2002) integrating cre-ativity and unexpected expansions of the original requirements This precludes onto-logical definitions in which the problem space is completely bounded a priori Inconsequence a degree of openness is necessary to integrate different kinds of detailin description from fully described ones to others with shallower semantics eg someproviding only references to generic assumptions
Theoretical frameworksWe will make the distinction that guidelines (as further defined) are the propositions thatdirectly predicate about the validity or appropriateness of learning design parts whileother theoretical propositions about learning that are not directly saying anythingabout the design of digital learning resources are considered as assumptions followingthe general sense provided in the Merriam-Webster dictionary lsquoa fact or statement (asa proposition axiom postulate or notion) taken for grantedrsquo
The fact that some assumption hypotheses or proposed law about learning is sustainedby a broader empirical base or is better sustained by rational means is out of the scopeof this work because here we focus on design practice (even though it is of an obviousinterest for researchers) In principle even unorthodox positions may be sustainedbecause the mechanism of microtheories and their inclusion mechanism allows forgreat flexibility in modelling them
Table 2 provides some examples of theoretical assumptions and related guidelines forlsquoextremersquo archetypical standpoints on learning These are not intended to reflect actualcommon positions but they are provided as illustrative examples of radical positionsthat could be modelled with our ontological schema
The examples provided in Table 2 go a step further than toolkits in Conole et al (2004)in making the design rationale explicit Nonetheless in many cases lsquocommon sensersquo isapplied to design and a shallower approach to specification could be used instead egproviding simply links to the theoretical standpoints used It should be noted also thatthe resulting guidelines vary greatly in their level of concreteness and some of themmay be considered as mandatory from some standpoints while others are mere generalrepresentations This would lead to different computational strategies
Guidelines and rulesThe WordNet (httpwordnetprincetonedu) thesaurus defines guidelines as lsquoguidancerelative to setting standards or determining a course of actionrsquo or lsquoa rule or principlethat provides guidance to appropriate behaviorrsquo In our case guidelines are statementsthat serve as criteria for decision in designing learning Given that e-learning designis still far from being a mature discipline guidelinesmdashas instances of recordedknowledgemdashcan be characterised as oc_NormativeSpecifications becausethey describe how learning resource design activities and their outcomes lsquoshould bersquoin a propositional form Propositions lsquohave some truth value in some context or
Semantic learning designs 343
copy 2006 The Author Journal compilation copy 2006 British Educational Communications and Technology Agency
ldquoworldrdquo rsquo Furthermore propositions lsquoare assumed to be representable (at least in prin-ciple) by a sentence in some formal or natural languagersquo This raises the need for alanguage that expresses guidelines in a general sense There are two main options forsuch representation be them formal logical sentences or reified guidelines
Formal logical sentence representation entails the formalisation through sentences ofthe constraints of the designs For example if lsquoevery learning resource must be activersquoaccording to some theoretical standpoint (eg one that fosters reflective learning) thissentence may be represented as an axiom and put into the appropriate separate micro-theory as the following CycL fragment
($forAll X ($implies
($isa X $LearningObject)
($resourceType X $Active))
Then resources that are not active are simply excluded from the notion itself This is astrong position in the sense that precludes some views on the learning design process
Table 2 Some examples of theoretical assumptions and guidelines
Archetypicalstandpoint Assumptions Guidelines
Instructivist(Behaviouristic)
bull Learning occurs through association and reinforcement
bull The results of learning can be appreciated through observable outcomes
(GB1) Every instance of oc-Learning entails theexposure to some LearningObjectthat produces the association (GB2) Learning Activitysequences should provide assessmentactivities that confirm theassociations that should have beenestablished so far
Constructivist bull Learning proceeds by building mental structures
(GC1) Agentrsquos mental structuresshould be represented(GC2) These structures should bepre-assessed in order to becomparable with the resulting ones(GC3) Design Activities arerecommended(GC4) Contrast Activities arerecommended
Sociocultural bull Learning occurs through social participatio
(GSC1) Every instance ofoc-Learning entails theinteraction of severaloc-IntelligentAgents asoc-Actors
344 British Journal of Educational Technology Vol 37 No 3 2006
copy 2006 The Author Journal compilation copy 2006 British Educational Communications and Technology Agency
Another typical example may be that of a sociocultural guideline that preclude Methodsthat do not entail multirole activities that provide access to social interaction ServicesThis approach to guidelines provides the best framework for automation and wouldenable the creation of microtheories that represent clearly formulated views onlearning Nonetheless there are two major problems for the use of this option as theunique mechanism for encoding design rationales
1 There is no clear-cut distinction between models and theories of learning becausethey come from a history of schools of thought with interrelations (Palmer 2001)This would lead to the need for the representation of dozens of variants reflectingpositions that only diverge in some aspects
2 Many guidelines cannot be represented as logical sentences because theyrepresent lsquorecommendationsrsquo or vague requirements eg see (GC3) or (GC4)(Table 2)
These issues suggest that a combination of formal sentences with reified guidelinesprovide the best option for flexibility The representation of guidelines has beenstudied in the field of HumanndashComputer-Interaction (HCI) in the last decade(Vanderdonckt 1999) and the main philosophy of such representations remainsvalid in the learning design domain The major challenge for such representations isthat of providing computational semantics that could be used for the implementationof software support but this is difficult to attain in all the cases Thus the concept oflearning design guideline will provide two kinds of representations general andschema-specific
The general use of instances of oc_NormativeSpecification allows the model-ling of lsquoconceptual works that describe how something should bersquo This may includerepresentation of known normative representations or procedures For example onemay refer to a concrete learning design model as that derived from Bannister (1981)which states that creativity cycles follow an initial phase of brainstorming (idea cap-ture) followed by a phase in which key issues are focused into an operation strategy(idea development) and a third phase in which project control takes place through alsquorecursive cyclersquo (operational management) The reference could be made by pointing tothe academic reference describing it as a conceptual work or it could be described asan Activity-Spec modelled explicitly inside the ontology As another examplesome detailed guidelines could be modelled by the normative specification sub-category of oc-Instructions defined as lsquoa sequence of tasks to be performed byan agentrsquo
Schema-specific guidelines restrict the freedom to represent criteria as conceptual workinstances to a concrete and commonly agreed ontology that determine what can be saidabout how learning designs should be according to a theoretical standpoint or frame-work These build on an activityndashrole ontology as the one described in the previoussection and add a prescription on the kind of lsquoshould-bersquo predications that are allow-able Table 3 describes examples of a minimal language that could be used for thatpurpose
Semantic learning designs 345
copy 2006 The Author Journal compilation copy 2006 British Educational Communications and Technology Agency
In the examples CAT represents a category of individual expressed through a conceptor a logical formula and the sentence is a synthetic way of expressing some recom-mended or mandatory constraint over such category Other languages could be engi-neered in further research but the one described can be directly expressed in OpenCycand combined with software modules that check or enforce them in some way Notethat this representation does not use direct OpenCyc sentences so that reification is usedinstead of direct logical sentences It is not the aim of this paper to determine the mostpowerful and flexible representation language but only set a foundation for furtherspecification efforts
Analysis and comparison of two theoretical semantic frameworksAs an assessment of the representational capabilities of the scheme just described acase study of representation is sketched in this section We will use the learning designtoolkit described by Conole et al (2004) as the underlying terminology even thoughother frameworks could be used instead
Two rather diverging positions for design will be used as an illustration Table 4 sum-marises the two design scenarios
Table 3 Possible language for use in learning designs
Sentence type Potential uses Example
RecommendationshouldBe (CATSENT)
Recommend kinds ofactivities
Design activities are recommended (GC3)shouldBe(Method subEvent (CATDesignActivity))
ObligationmustBe (CATSENT)
Some activity typesmust follow others
Assessments must follow each expositionto contents (GD1) mustBe(and(ExpositiveActivityIndividualActivity)startAfterEndingOf(CAT
AssessmentActivity)
Activities must havesome characteristic
Every activity must entail interactionmustBe(LDActivity(usedIn(CAT
InteractiveService)))Every activity must provide acontent-oriented learning object forindividual use (similar to GB1)mustBe(LDActivity(and (usedIn(CAT
LearningObject)(learnerRoleNumber(1))))
Some structure shouldbe enforced in methods
Methods should have a sequential actstructuremustBe(Methodnot ((subEvent(CAT Play))))
CAT = category GC3 = Constructivist Guideline 3 (see Table 2) GD1 = Guideline 1
346 British Journal of Educational Technology Vol 37 No 3 2006
copy 2006 The Author Journal compilation copy 2006 British Educational Communications and Technology Agency
The following subsections sketch the main outcomes of an example learning designprocess for both scenarios assuming that the process described by Conole et al (2004)is used in both situations
Differences in modelling the design processThe forces driving the design process in both scenarios are fairly diverging The designof Scenario A will typically be constrained by a clear target specified in terms of anaggregated competency level while in Scenario B there are no concrete competenciesto be acquired and the main objective of the design activity would be twofold On theone hand the activities are targeted at promoting interchange of ideas in an attemptto stimulate casual convergence of interest which is the main value of the culture ofthe special interest group On the other hand the activities have the generic objectiveof strengthening social ties which is an important element of cohesion for the under-taking of joint projects
Table 5 describes the main inputs and possible activities and their objectives at a highlevel
The changes intended to be achieved in each of the scenarios are stated differently witha formal definition in (A) and a shallow cognitive and social representation in the case
Table 4 Learning design scenarios
Scenario DescriptionClassification according to
Conole et al (2004) toolkit
(A) Systemic-Behaviourist A software development company attempts to designlearning units about newtechnologies or softwareproducts for its developerstaff The design is drivenby a cycle of lsquocompetencygap analysisrsquo that modelsorganisational learning in terms of the aggregated competencies of the employeesmdashsimilar to theone described by Sicilia(2005)
Focuses on the individual where a body of informationforms the basis and rawmaterial for learning Learning is mainly considered as nonreflective ie based on skill learning and memorisation
(B) Sociocultural A special interest group ona concrete research topicaims at developing the knowledge of its membersby sharing experiencesand discussing on themesof common interest
Focuses on the social whereexperience is the main source of learning Conscious reflection on experience is transformed into learning
Semantic learning designs 347
copy 2006 The Author Journal compilation copy 2006 British Educational Communications and Technology Agency
of (B) In addition case (B) requires a model of social network that is largely irrelevantin case (A)
Differences in recording activity rationalesThe approach for the recording of design rationales also differs because of the differentcontexts Table 6 summarises the main elements for each of the scenarios
Conclusions and directions for researchCurrent technology and practice related to the description of reusable digital learningresources has resulted in techniques that allow the description of learning activities of
Table 5 Inputs activities and objectives of two learning design scenarios
Scenario Design inputs Activities and their objectives
(A) Systemic-Behaviourist A formal competency recordfor employees and competencyrequirements expressed in terms of amounts of required competencies (assuming a shared measurement scale) Such requirements are actually derived from strategy or from the contingent requirements of upcoming projects
Competency instances are used to measure the levelof Competency-Spec for the required competencies (or their constituents) A formal match considering the agenda of employees is carried out for determining the target learners LearningObjects forindividual use are selected asthe main elements of theactivities and the criteria forselection are again a match ofthe objectives stated inmetadata (as stated incategory 9 of Learning Object MetadatamdashClassification) and thecompetencies required
(B) Sociocultural The expertise of the special interest group members as stated in the concepts addressed by their publicationsprojects or other activities
Informal group activities are the main approach applicablesince there is not a predefinedgroup structure but a cultureof sharing and socialisation The results of the sharing of ideas can be expressed in terms of oc-knowsAboutpredicates and social ties will be modelled through positive valence FOAF-knows(httpxmlnscomfoaf01)predicates with an added numerical value representingstrength
348 British Journal of Educational Technology Vol 37 No 3 2006
copy 2006 The Author Journal compilation copy 2006 British Educational Communications and Technology Agency
a complex kind including their intended participants educational objectives and exe-cution constraints The representation and recording of the assumptions and guidelinesthat were considered in the creation of such resources would extend current techniquesto capture the rationale of the design itself This in turn will eventually result in theavailability of a large base of empirical data available in which theoretical assumptionsor rules are explicitly connected to the resulting designs as these could be linked withthe actual results and trace of the learning processes that used them The resulting bodyof evidence would represent a critical asset in the development of the theory of learningmediated through computers
This paper has described the first attempt to extend existing learning technology modelswith a schema explicitly targeted at describing learning design rationales The OpenCycknowledge base has been used as the semantic reference from which the ontologicalschema has been specified Concretely an activity-based schema based on IMS LD hasbeen described and theoretical assumptions and codified guidelines have been intro-duced The flexibility of the representation schema has been analysed through therepresentation of two stereotyped theoretical positions
Two main research directions should continue the work described here On the onehand further ontological engineering effort is required to come up with a more com-prehensive and rich schema for learning design rationales On the other hand tool
Table 6 Main elements of the two learning design scenarios
Scenario Approach Examples
(A) Systemic-Behaviourist Formal logical sentencesare used to guarantee that every design followsan organisationalobjective
It is required that the objective of every Method is attached to a Competency-Spec instance as objective
Formal assessment is a requirement foraccountability purposes
Guideline (GD1) is attached to everyDesigningLearningResource instance that produces an Act
(B) Sociocultural The interaction in informal settings can beused to derive social models (Sicilia amp Garciacutea2004)
The interactions in discussion canbe used to measure the degree of lsquoproximityrsquo between members Therationale of the activities can bejustified in terms of SocialTies inside SocialNetworks Further the focus of each groupactivity should be clearly stated anda combination of the increase in collective knowledge and the strengthening of social relations can be used for assessment
Semantic learning designs 349
copy 2006 The Author Journal compilation copy 2006 British Educational Communications and Technology Agency
implementation and record gathering efforts would be required to attain the ultimateobjective of the ontological definitions described so far With such effort an unprece-dented research material could be available in some years providing improved oppor-tunities for scientific inquiry on design as an activity regarding pedagogy
Other potential directions for research may be the description of interpretations of thedesign intentions (for which no rationale was explicitly recorded) the mining or detec-tion of patterns as related to rationale description or the detailed use of theoreticalrationales in agents that support the learning process which represent a category ofrelevant learning support software (Holmes in press)
ReferencesAllert H (2004) Coherent social systems for learningmdashan approach for contextualized and
community-centred metadata Journal of Interactive Media in Education 2004 1Baader F Calvanese D McGuinness D Nardi D amp Patel-Schneider P (Eds) (2003) The
description logic handbook Theory implementation and applications Cambridge CambridgeUniversity Press
Bannister D (1981) Personal construct theory and research method In P Reason amp J Rowan(Eds) Human inquiry a sourcebook of new paradigm research (pp 191ndash200) Chichester UKJohn Wiley and Sons
Berners-Lee T Hendler J amp Lassila O (2001) The semantic web Scientific American 284 534ndash43
Brown A amp Voltz B (2005) Elements of effective e-learning design International Review ofResearch in Open and Distance Learning 6 1
Buzza D Bean D Harrigan K amp Carey T (2005) Learning design repositories adaptinglearning design specifications for shared instructional knowledge Canadian Journal of Learningand Technology 30 3
Conole G Dyke M Oliver M amp Seale J (2004) Mapping pedagogy and tools for effectivelearning design Computers amp Education 43 1ndash2 17ndash33
Dodero J M Aedo I amp Diacuteaz P (2002) Participative knowledge production of learning objectsfor e-books The Electronic Library 20 4 296ndash305
Friesen N (2005) Interoperability and learning objects an overview of e-learning standardiza-tion Interdisciplinary Journal of Knowledge and Learning Objects 1 1 23ndash31
Gruber T (1995) Towards principles for the design of ontologies used for knowledge sharingInternational Journal of HumanndashComputer Studies 43 56 907ndash928
Hatchuel A (2002) Towards design theory and expandable rationality the unfinished programof Herbert Simon Journal of Management and Governance 5 3ndash4 260ndash271
Holmes J (in press) Designing agents to support learning by explaining Computers and EducationIEEE Learning Technology Standards Committee (2002) Learning object metadata (LOM) Final
Draft Standard IEEE 1484121ndash2002IMS (2003) IMS Learning Design Information Model (Version 10) Final Specification 20
January 2003Kaptelinin V Kuutti K amp Bannon L (1995) Activity theory basic concepts and applications
In Blumenthalet al (Eds) Humanndashcomputer interaction Lecture Notes in Computer Science 1015189ndash201
Koper R (2004) Use of the semantic web to solve some basic problems in education increaseflexible distributed lifelong learning decrease teacherrsquos workload Journal of Interactive Mediain Education 2004 6
Lenat D (1995) CYC a large-scale investment in knowledge infrastructure Communications ofthe ACM 38 11 33ndash38
350 British Journal of Educational Technology Vol 37 No 3 2006
copy 2006 The Author Journal compilation copy 2006 British Educational Communications and Technology Agency
McGreal R (2004) Learning objects a practical definition International Journal of InstructionalTechnology and Distance Learning 1 9 21ndash32
Packer M amp Goicoechea J (2000) Sociocultural and constructivist theories of learningontology not just epistemology Educational Psychologist 35 4 227ndash241
Palmer J (2001) Fifty major thinkers on education London and New York RoutledgePolsani P R (2003) Use and abuse of reusable learning objects Journal of Digital Information 3
4 Rothwell W amp Kazanas H (1992) Mastering the instructional design process San Francisco CA
Jossey-BassSaacutenchez-Alonso S amp Sicilia M A (2004) Relationships and commitments in learning object
metadata In Proceedings of the 5th International Conference on Information Technology BasedHigher Education and Training ITHET 2004 Istanbul Turkey
Sicilia M A (2005) Ontology-Based Competency Management Infrastructures for theKnowledge-intensive Learning Organization In Lytras amp Naeve (Eds) Intelligent Learning Infra-structures in Knowledge Intensive Organizations A Semantic Web perspective IDEA USA pp 302ndash324
Sicilia M A amp Garciacutea E (2004) Fuzzy group models for adaptation in cooperative informationretrieval contexts Lecture Notes in Computer Science 2932 324ndash334
Sicilia M A amp Garciacutea E (2005) On the convergence of formal ontologies and standardizede-learning Journal of Distance Education Technologies 3 2 13ndash29
Sicilia M A amp Lytras M (2005) On the representation of change according to different ontol-ogies of learning International Journal of Learning and Change 1 1 66ndash79
Sicilia M A Lytras M Rodriacuteguez E amp Garciacutea E (in press) Integrating descriptions of knowl-edge management learning activities into large ontological structures a case study Data andKnowledge Engineering
Tennant M (1991) Expertise as a dimension of adult development New Education 13 1 49ndash55Vanderdonckt J (1999) Development milestones towards a tool for working with guidelines
Interacting with Computers 12 2 81ndash118Wand Y amp Weber R (1995) On the deep structure of information systems Information Systems
Journal 5 203ndash223Wiley D A (2001) The instructional use of learning objects Bloomington IN Association for
Educational Communications and Technology
338 British Journal of Educational Technology Vol 37 No 3 2006
copy 2006 The Author Journal compilation copy 2006 British Educational Communications and Technology Agency
Figure 1 (a) Summary of the competency components in the ontological schema (b) Summary of the main activityndashrole definitions in the ontological schema
oc_WorkingEvent
oc_IntelligentAgent
The competency elements aremeasurable characteristics andthey will be typed by the relatedSpecification classes eg Skill
SpecputlnPracticeln
(a)
Competency Competency-Spec
Competencyelement
Skill oc_ Competence
Attitude
Knowledge
has
thingSpecified
oc_Learning
ComputerMediatedLearning
oc_Role
(b)
Staff
Learner
oc_Specification
Method-Spec
Play-Spec
Act-Spec
Support
LD-Activity
contains
contains
Method
Play
Act
startAfterEndingOf
subAction
subAction
Activity-Spec
ltltActorSlotsgtgt
situationConformsToSpec
Activity-Spec
Activity
Semantic learning designs 339
copy 2006 The Author Journal compilation copy 2006 British Educational Communications and Technology Agency
The concept of oc_Role accounts not only for participant roles but also for otherkinds of lsquocomponentsrsquo This is why two specialisations of the concept ofoc_ActorSlot labelled Staff and Learner are required to differentiate partici-pants from other elements As is obvious from the earlier discussion the built-in rep-resentation of activities in OpenCyc is powerful enough for the definition of learningdesign activities
The final part of the ontological specification described so far is that of the mappingfrom LD specifications to the actual past or present activity realisations Theoc_situationConformsToSpecification predicate can be used to trace whichactivity realizations come from a given learning design activity definition and the sameoccurs with the oc_thingSpecified for other learning design elements such asroles or environments Existing learning design execution engines as CopperCore(httpcoppercoreorg) actually map LD specifications to online collaboration so thatthe translation of the ontological representations of learning designs to ontology defi-nitions of activities is straightforward following similar representation
Describing objectives outcomes and preconditionsAccording to IMS learning design learning objectives and prerequisites lsquocan bedescribed using the IMS Reusable Definition of Competency or Educational Objective(RDCEO) format but can also refer to simple resources (eg a text) with a descriptionof the learning objectiversquo As the latter option is not useful from the viewpoint of com-putational semantics we will adhere to the former As commented earlier the compe-tency view is not exclusive to other kinds of objectives related to other ontologicalstandpoints (Sicilia amp Lytras 2005) and other conceptions of change or improvementmay include social ties cognitive structures or the combination of the knowledge of agroup In fact competencies adhere mostly to behaviouristic frameworks that focus onobservable external behaviour (Tennant 1991) rather than on cognitive representa-tions Nonetheless the further discussion is equally applicable to other kinds of objec-tives so that we will adhere to the competency view
oc_Competence in OpenCyc is defined as a quantity that lsquofor describing the level ofskill with which an agent performs some taskrsquo This does not remove the lack of preci-sion that is yet present in RDCEO definitions as described by Sicilia (2005) As analternative Figure 1a provides a simplified account of the model described by Sicilia asrelated to OpenCyc terms in which competence is integrated as one of the elements of thebroader notion of competency according to the model of Rummel (Rothwell amp Kazanas1992) which considers work situations attributes of the individual that are applied insuch situations and measurement with respect to some standard of performance Themost relevant distinction is again the separation of actual competency representationsand competency lsquodefinitionsrsquo (ie specifications) intended to provide a type structure tocompetencies to be used in objective or precondition specification In this case thegeneric oc-thingSpecified predicate can be used for the linking of both levels ofrepresentation
340 British Journal of Educational Technology Vol 37 No 3 2006
copy 2006 The Author Journal compilation copy 2006 British Educational Communications and Technology Agency
Objectives and preconditions can then be expressed as formulas on the acquisition ofcompetencies or their components (competency elements) Descriptions specific toOpenCyc may use oc_CycLSentence-Assertible but simpler enumeration ofcompetency components could be used instead
Describing learning design rationales theories hypotheses guidelines and rulesThe concept of lsquodesignrsquo for our purposes is properly captured in the 5a definitionprovided in the Merriam Webster Online Dictionary (wwwm-wcom) lsquoan underlyingscheme that governs functioning developing or unfoldingrsquo In consequence learningdesigns refer to artefacts and not to the process of devising the educational solutionThis is equivalent to the concept oc_Design subsumed by the notion ofoc_Specification which is described as lsquoan abstract work that constitutes adescription of the properties of a Situation or a SomethingExisting and sometimeseven entire collections of such thingsrsquo In consequence the term LearningDesignwould simply delineate a subset of specifications that are oriented to describe howlearning experiences with concrete objectives should be Trivially oc_Designerdescribes the authors of the learning designs and oc_Designing represent theoc_HumanActivity that leads to the production of learning designs Designing as aspecial kind of activity is carried out (oc_doneBy) by a designer which provides a wayto attach metainformation about the rationale of the design
From an ontological perspective the departing problem in modelling learning designrationales is that their assumptions and hypotheses of different theoretical standpointsneed not be compatible or need not consider the same aspects of reality (Conole et al2004 Packer amp Goicoechea 2000) Furthermore different pedagogical approaches areknown to result in different mappings to activities (Conole et al 2004) This wouldeither require the provision of separate ontologies or the use of a representationalmechanism that allows such kind of potential inconsistency or divergence The conceptof microtheory in OpenCyc provides such representational mechanism intended to orga-nise assertions that depend on lsquoshared set of assumptions on which the truth of theassertions dependsrsquo Definitions inside the same microtheory need to be consistent butthis is not required across microtheories The descriptions provided in the earlier sectioncan be used as a base microtheory called LearningActivitiesMt for others that are spe-cific to concrete theoretical frameworks In this section we will describe elements thatare common to the description of theoretical frameworks the distinction among micro-theories will not be discussed
The intellectual processes of learning design in themselves will be represented asoc_Designing activities carried out by oc_Designers An additional consider-ation is that the information about the rationale of the design would be better attachedto representations of each of the design processes (ie to oc_Designing instances)than to the final design This enables the registering of different assumptions for differ-ent contributions to the same artefact eg different revisions from the same of differentdesigners
Semantic learning designs 341
copy 2006 The Author Journal compilation copy 2006 British Educational Communications and Technology Agency
Several designers may participate in the same designing activity and formal frameworksfor collaborative learning resource creation (Dodero Aedo amp Diacuteaz 2002) can be inte-grated by clearly specifying which class of things can be the outcome of a designingprocess For the sake of flexibility the concept of LearningDesignPart is introduced as anall-encompassing category that includes each of the elements described in the previoussection This includes not only activities and roles but also objectives conditions andthe like Furthermore the granularity of the descriptions allows for different levels fromconsidering a unique designing activity for the whole package to recording each ele-ment of the learning design comprehensively Because oc_Actions are subsumed bythe concept of oc_Event considered as a lsquodynamic situation in which the state of theworld changesrsquo each edition of a metadata record or any other form of description maybe subject to be described The concept is flexible enough to be able to record fine-grained detailed decisions This of course may include different versions of the sameartefacts but this is an orthogonal aspect of configuration that is not relevant to ourpresent discussion
Figure 2 depicts the main elements of the loci to which rationales assumptions andguidelines will be attached
A LearningDesigner can be defined as the class of humans that performactivities of learning resource design The consideration of learning design as aunique human activity could be considered as restrictive of intelligent softwarefor the task Simply changing the oc_Designer class to a subconcept ofoc_IntelligentAgent could remove this but this entails a change in Open-Cyc knowledge base The concept oc_Designing is a subclass of the conceptoc_ThinkingAboutTheStateOfTheWorld which represents lsquothinking pro-cessesrsquo which include mental events such as planning evaluating or inventingThus the concept of design described so far should be understood in terms of the
Figure 2 Main elements of the ontological schema for the loci of design rationales
oc_performedBy modifies
342 British Journal of Educational Technology Vol 37 No 3 2006
copy 2006 The Author Journal compilation copy 2006 British Educational Communications and Technology Agency
concept of lsquoexpanded rationalityrsquo as described by Hatchuel (2002) integrating cre-ativity and unexpected expansions of the original requirements This precludes onto-logical definitions in which the problem space is completely bounded a priori Inconsequence a degree of openness is necessary to integrate different kinds of detailin description from fully described ones to others with shallower semantics eg someproviding only references to generic assumptions
Theoretical frameworksWe will make the distinction that guidelines (as further defined) are the propositions thatdirectly predicate about the validity or appropriateness of learning design parts whileother theoretical propositions about learning that are not directly saying anythingabout the design of digital learning resources are considered as assumptions followingthe general sense provided in the Merriam-Webster dictionary lsquoa fact or statement (asa proposition axiom postulate or notion) taken for grantedrsquo
The fact that some assumption hypotheses or proposed law about learning is sustainedby a broader empirical base or is better sustained by rational means is out of the scopeof this work because here we focus on design practice (even though it is of an obviousinterest for researchers) In principle even unorthodox positions may be sustainedbecause the mechanism of microtheories and their inclusion mechanism allows forgreat flexibility in modelling them
Table 2 provides some examples of theoretical assumptions and related guidelines forlsquoextremersquo archetypical standpoints on learning These are not intended to reflect actualcommon positions but they are provided as illustrative examples of radical positionsthat could be modelled with our ontological schema
The examples provided in Table 2 go a step further than toolkits in Conole et al (2004)in making the design rationale explicit Nonetheless in many cases lsquocommon sensersquo isapplied to design and a shallower approach to specification could be used instead egproviding simply links to the theoretical standpoints used It should be noted also thatthe resulting guidelines vary greatly in their level of concreteness and some of themmay be considered as mandatory from some standpoints while others are mere generalrepresentations This would lead to different computational strategies
Guidelines and rulesThe WordNet (httpwordnetprincetonedu) thesaurus defines guidelines as lsquoguidancerelative to setting standards or determining a course of actionrsquo or lsquoa rule or principlethat provides guidance to appropriate behaviorrsquo In our case guidelines are statementsthat serve as criteria for decision in designing learning Given that e-learning designis still far from being a mature discipline guidelinesmdashas instances of recordedknowledgemdashcan be characterised as oc_NormativeSpecifications becausethey describe how learning resource design activities and their outcomes lsquoshould bersquoin a propositional form Propositions lsquohave some truth value in some context or
Semantic learning designs 343
copy 2006 The Author Journal compilation copy 2006 British Educational Communications and Technology Agency
ldquoworldrdquo rsquo Furthermore propositions lsquoare assumed to be representable (at least in prin-ciple) by a sentence in some formal or natural languagersquo This raises the need for alanguage that expresses guidelines in a general sense There are two main options forsuch representation be them formal logical sentences or reified guidelines
Formal logical sentence representation entails the formalisation through sentences ofthe constraints of the designs For example if lsquoevery learning resource must be activersquoaccording to some theoretical standpoint (eg one that fosters reflective learning) thissentence may be represented as an axiom and put into the appropriate separate micro-theory as the following CycL fragment
($forAll X ($implies
($isa X $LearningObject)
($resourceType X $Active))
Then resources that are not active are simply excluded from the notion itself This is astrong position in the sense that precludes some views on the learning design process
Table 2 Some examples of theoretical assumptions and guidelines
Archetypicalstandpoint Assumptions Guidelines
Instructivist(Behaviouristic)
bull Learning occurs through association and reinforcement
bull The results of learning can be appreciated through observable outcomes
(GB1) Every instance of oc-Learning entails theexposure to some LearningObjectthat produces the association (GB2) Learning Activitysequences should provide assessmentactivities that confirm theassociations that should have beenestablished so far
Constructivist bull Learning proceeds by building mental structures
(GC1) Agentrsquos mental structuresshould be represented(GC2) These structures should bepre-assessed in order to becomparable with the resulting ones(GC3) Design Activities arerecommended(GC4) Contrast Activities arerecommended
Sociocultural bull Learning occurs through social participatio
(GSC1) Every instance ofoc-Learning entails theinteraction of severaloc-IntelligentAgents asoc-Actors
344 British Journal of Educational Technology Vol 37 No 3 2006
copy 2006 The Author Journal compilation copy 2006 British Educational Communications and Technology Agency
Another typical example may be that of a sociocultural guideline that preclude Methodsthat do not entail multirole activities that provide access to social interaction ServicesThis approach to guidelines provides the best framework for automation and wouldenable the creation of microtheories that represent clearly formulated views onlearning Nonetheless there are two major problems for the use of this option as theunique mechanism for encoding design rationales
1 There is no clear-cut distinction between models and theories of learning becausethey come from a history of schools of thought with interrelations (Palmer 2001)This would lead to the need for the representation of dozens of variants reflectingpositions that only diverge in some aspects
2 Many guidelines cannot be represented as logical sentences because theyrepresent lsquorecommendationsrsquo or vague requirements eg see (GC3) or (GC4)(Table 2)
These issues suggest that a combination of formal sentences with reified guidelinesprovide the best option for flexibility The representation of guidelines has beenstudied in the field of HumanndashComputer-Interaction (HCI) in the last decade(Vanderdonckt 1999) and the main philosophy of such representations remainsvalid in the learning design domain The major challenge for such representations isthat of providing computational semantics that could be used for the implementationof software support but this is difficult to attain in all the cases Thus the concept oflearning design guideline will provide two kinds of representations general andschema-specific
The general use of instances of oc_NormativeSpecification allows the model-ling of lsquoconceptual works that describe how something should bersquo This may includerepresentation of known normative representations or procedures For example onemay refer to a concrete learning design model as that derived from Bannister (1981)which states that creativity cycles follow an initial phase of brainstorming (idea cap-ture) followed by a phase in which key issues are focused into an operation strategy(idea development) and a third phase in which project control takes place through alsquorecursive cyclersquo (operational management) The reference could be made by pointing tothe academic reference describing it as a conceptual work or it could be described asan Activity-Spec modelled explicitly inside the ontology As another examplesome detailed guidelines could be modelled by the normative specification sub-category of oc-Instructions defined as lsquoa sequence of tasks to be performed byan agentrsquo
Schema-specific guidelines restrict the freedom to represent criteria as conceptual workinstances to a concrete and commonly agreed ontology that determine what can be saidabout how learning designs should be according to a theoretical standpoint or frame-work These build on an activityndashrole ontology as the one described in the previoussection and add a prescription on the kind of lsquoshould-bersquo predications that are allow-able Table 3 describes examples of a minimal language that could be used for thatpurpose
Semantic learning designs 345
copy 2006 The Author Journal compilation copy 2006 British Educational Communications and Technology Agency
In the examples CAT represents a category of individual expressed through a conceptor a logical formula and the sentence is a synthetic way of expressing some recom-mended or mandatory constraint over such category Other languages could be engi-neered in further research but the one described can be directly expressed in OpenCycand combined with software modules that check or enforce them in some way Notethat this representation does not use direct OpenCyc sentences so that reification is usedinstead of direct logical sentences It is not the aim of this paper to determine the mostpowerful and flexible representation language but only set a foundation for furtherspecification efforts
Analysis and comparison of two theoretical semantic frameworksAs an assessment of the representational capabilities of the scheme just described acase study of representation is sketched in this section We will use the learning designtoolkit described by Conole et al (2004) as the underlying terminology even thoughother frameworks could be used instead
Two rather diverging positions for design will be used as an illustration Table 4 sum-marises the two design scenarios
Table 3 Possible language for use in learning designs
Sentence type Potential uses Example
RecommendationshouldBe (CATSENT)
Recommend kinds ofactivities
Design activities are recommended (GC3)shouldBe(Method subEvent (CATDesignActivity))
ObligationmustBe (CATSENT)
Some activity typesmust follow others
Assessments must follow each expositionto contents (GD1) mustBe(and(ExpositiveActivityIndividualActivity)startAfterEndingOf(CAT
AssessmentActivity)
Activities must havesome characteristic
Every activity must entail interactionmustBe(LDActivity(usedIn(CAT
InteractiveService)))Every activity must provide acontent-oriented learning object forindividual use (similar to GB1)mustBe(LDActivity(and (usedIn(CAT
LearningObject)(learnerRoleNumber(1))))
Some structure shouldbe enforced in methods
Methods should have a sequential actstructuremustBe(Methodnot ((subEvent(CAT Play))))
CAT = category GC3 = Constructivist Guideline 3 (see Table 2) GD1 = Guideline 1
346 British Journal of Educational Technology Vol 37 No 3 2006
copy 2006 The Author Journal compilation copy 2006 British Educational Communications and Technology Agency
The following subsections sketch the main outcomes of an example learning designprocess for both scenarios assuming that the process described by Conole et al (2004)is used in both situations
Differences in modelling the design processThe forces driving the design process in both scenarios are fairly diverging The designof Scenario A will typically be constrained by a clear target specified in terms of anaggregated competency level while in Scenario B there are no concrete competenciesto be acquired and the main objective of the design activity would be twofold On theone hand the activities are targeted at promoting interchange of ideas in an attemptto stimulate casual convergence of interest which is the main value of the culture ofthe special interest group On the other hand the activities have the generic objectiveof strengthening social ties which is an important element of cohesion for the under-taking of joint projects
Table 5 describes the main inputs and possible activities and their objectives at a highlevel
The changes intended to be achieved in each of the scenarios are stated differently witha formal definition in (A) and a shallow cognitive and social representation in the case
Table 4 Learning design scenarios
Scenario DescriptionClassification according to
Conole et al (2004) toolkit
(A) Systemic-Behaviourist A software development company attempts to designlearning units about newtechnologies or softwareproducts for its developerstaff The design is drivenby a cycle of lsquocompetencygap analysisrsquo that modelsorganisational learning in terms of the aggregated competencies of the employeesmdashsimilar to theone described by Sicilia(2005)
Focuses on the individual where a body of informationforms the basis and rawmaterial for learning Learning is mainly considered as nonreflective ie based on skill learning and memorisation
(B) Sociocultural A special interest group ona concrete research topicaims at developing the knowledge of its membersby sharing experiencesand discussing on themesof common interest
Focuses on the social whereexperience is the main source of learning Conscious reflection on experience is transformed into learning
Semantic learning designs 347
copy 2006 The Author Journal compilation copy 2006 British Educational Communications and Technology Agency
of (B) In addition case (B) requires a model of social network that is largely irrelevantin case (A)
Differences in recording activity rationalesThe approach for the recording of design rationales also differs because of the differentcontexts Table 6 summarises the main elements for each of the scenarios
Conclusions and directions for researchCurrent technology and practice related to the description of reusable digital learningresources has resulted in techniques that allow the description of learning activities of
Table 5 Inputs activities and objectives of two learning design scenarios
Scenario Design inputs Activities and their objectives
(A) Systemic-Behaviourist A formal competency recordfor employees and competencyrequirements expressed in terms of amounts of required competencies (assuming a shared measurement scale) Such requirements are actually derived from strategy or from the contingent requirements of upcoming projects
Competency instances are used to measure the levelof Competency-Spec for the required competencies (or their constituents) A formal match considering the agenda of employees is carried out for determining the target learners LearningObjects forindividual use are selected asthe main elements of theactivities and the criteria forselection are again a match ofthe objectives stated inmetadata (as stated incategory 9 of Learning Object MetadatamdashClassification) and thecompetencies required
(B) Sociocultural The expertise of the special interest group members as stated in the concepts addressed by their publicationsprojects or other activities
Informal group activities are the main approach applicablesince there is not a predefinedgroup structure but a cultureof sharing and socialisation The results of the sharing of ideas can be expressed in terms of oc-knowsAboutpredicates and social ties will be modelled through positive valence FOAF-knows(httpxmlnscomfoaf01)predicates with an added numerical value representingstrength
348 British Journal of Educational Technology Vol 37 No 3 2006
copy 2006 The Author Journal compilation copy 2006 British Educational Communications and Technology Agency
a complex kind including their intended participants educational objectives and exe-cution constraints The representation and recording of the assumptions and guidelinesthat were considered in the creation of such resources would extend current techniquesto capture the rationale of the design itself This in turn will eventually result in theavailability of a large base of empirical data available in which theoretical assumptionsor rules are explicitly connected to the resulting designs as these could be linked withthe actual results and trace of the learning processes that used them The resulting bodyof evidence would represent a critical asset in the development of the theory of learningmediated through computers
This paper has described the first attempt to extend existing learning technology modelswith a schema explicitly targeted at describing learning design rationales The OpenCycknowledge base has been used as the semantic reference from which the ontologicalschema has been specified Concretely an activity-based schema based on IMS LD hasbeen described and theoretical assumptions and codified guidelines have been intro-duced The flexibility of the representation schema has been analysed through therepresentation of two stereotyped theoretical positions
Two main research directions should continue the work described here On the onehand further ontological engineering effort is required to come up with a more com-prehensive and rich schema for learning design rationales On the other hand tool
Table 6 Main elements of the two learning design scenarios
Scenario Approach Examples
(A) Systemic-Behaviourist Formal logical sentencesare used to guarantee that every design followsan organisationalobjective
It is required that the objective of every Method is attached to a Competency-Spec instance as objective
Formal assessment is a requirement foraccountability purposes
Guideline (GD1) is attached to everyDesigningLearningResource instance that produces an Act
(B) Sociocultural The interaction in informal settings can beused to derive social models (Sicilia amp Garciacutea2004)
The interactions in discussion canbe used to measure the degree of lsquoproximityrsquo between members Therationale of the activities can bejustified in terms of SocialTies inside SocialNetworks Further the focus of each groupactivity should be clearly stated anda combination of the increase in collective knowledge and the strengthening of social relations can be used for assessment
Semantic learning designs 349
copy 2006 The Author Journal compilation copy 2006 British Educational Communications and Technology Agency
implementation and record gathering efforts would be required to attain the ultimateobjective of the ontological definitions described so far With such effort an unprece-dented research material could be available in some years providing improved oppor-tunities for scientific inquiry on design as an activity regarding pedagogy
Other potential directions for research may be the description of interpretations of thedesign intentions (for which no rationale was explicitly recorded) the mining or detec-tion of patterns as related to rationale description or the detailed use of theoreticalrationales in agents that support the learning process which represent a category ofrelevant learning support software (Holmes in press)
ReferencesAllert H (2004) Coherent social systems for learningmdashan approach for contextualized and
community-centred metadata Journal of Interactive Media in Education 2004 1Baader F Calvanese D McGuinness D Nardi D amp Patel-Schneider P (Eds) (2003) The
description logic handbook Theory implementation and applications Cambridge CambridgeUniversity Press
Bannister D (1981) Personal construct theory and research method In P Reason amp J Rowan(Eds) Human inquiry a sourcebook of new paradigm research (pp 191ndash200) Chichester UKJohn Wiley and Sons
Berners-Lee T Hendler J amp Lassila O (2001) The semantic web Scientific American 284 534ndash43
Brown A amp Voltz B (2005) Elements of effective e-learning design International Review ofResearch in Open and Distance Learning 6 1
Buzza D Bean D Harrigan K amp Carey T (2005) Learning design repositories adaptinglearning design specifications for shared instructional knowledge Canadian Journal of Learningand Technology 30 3
Conole G Dyke M Oliver M amp Seale J (2004) Mapping pedagogy and tools for effectivelearning design Computers amp Education 43 1ndash2 17ndash33
Dodero J M Aedo I amp Diacuteaz P (2002) Participative knowledge production of learning objectsfor e-books The Electronic Library 20 4 296ndash305
Friesen N (2005) Interoperability and learning objects an overview of e-learning standardiza-tion Interdisciplinary Journal of Knowledge and Learning Objects 1 1 23ndash31
Gruber T (1995) Towards principles for the design of ontologies used for knowledge sharingInternational Journal of HumanndashComputer Studies 43 56 907ndash928
Hatchuel A (2002) Towards design theory and expandable rationality the unfinished programof Herbert Simon Journal of Management and Governance 5 3ndash4 260ndash271
Holmes J (in press) Designing agents to support learning by explaining Computers and EducationIEEE Learning Technology Standards Committee (2002) Learning object metadata (LOM) Final
Draft Standard IEEE 1484121ndash2002IMS (2003) IMS Learning Design Information Model (Version 10) Final Specification 20
January 2003Kaptelinin V Kuutti K amp Bannon L (1995) Activity theory basic concepts and applications
In Blumenthalet al (Eds) Humanndashcomputer interaction Lecture Notes in Computer Science 1015189ndash201
Koper R (2004) Use of the semantic web to solve some basic problems in education increaseflexible distributed lifelong learning decrease teacherrsquos workload Journal of Interactive Mediain Education 2004 6
Lenat D (1995) CYC a large-scale investment in knowledge infrastructure Communications ofthe ACM 38 11 33ndash38
350 British Journal of Educational Technology Vol 37 No 3 2006
copy 2006 The Author Journal compilation copy 2006 British Educational Communications and Technology Agency
McGreal R (2004) Learning objects a practical definition International Journal of InstructionalTechnology and Distance Learning 1 9 21ndash32
Packer M amp Goicoechea J (2000) Sociocultural and constructivist theories of learningontology not just epistemology Educational Psychologist 35 4 227ndash241
Palmer J (2001) Fifty major thinkers on education London and New York RoutledgePolsani P R (2003) Use and abuse of reusable learning objects Journal of Digital Information 3
4 Rothwell W amp Kazanas H (1992) Mastering the instructional design process San Francisco CA
Jossey-BassSaacutenchez-Alonso S amp Sicilia M A (2004) Relationships and commitments in learning object
metadata In Proceedings of the 5th International Conference on Information Technology BasedHigher Education and Training ITHET 2004 Istanbul Turkey
Sicilia M A (2005) Ontology-Based Competency Management Infrastructures for theKnowledge-intensive Learning Organization In Lytras amp Naeve (Eds) Intelligent Learning Infra-structures in Knowledge Intensive Organizations A Semantic Web perspective IDEA USA pp 302ndash324
Sicilia M A amp Garciacutea E (2004) Fuzzy group models for adaptation in cooperative informationretrieval contexts Lecture Notes in Computer Science 2932 324ndash334
Sicilia M A amp Garciacutea E (2005) On the convergence of formal ontologies and standardizede-learning Journal of Distance Education Technologies 3 2 13ndash29
Sicilia M A amp Lytras M (2005) On the representation of change according to different ontol-ogies of learning International Journal of Learning and Change 1 1 66ndash79
Sicilia M A Lytras M Rodriacuteguez E amp Garciacutea E (in press) Integrating descriptions of knowl-edge management learning activities into large ontological structures a case study Data andKnowledge Engineering
Tennant M (1991) Expertise as a dimension of adult development New Education 13 1 49ndash55Vanderdonckt J (1999) Development milestones towards a tool for working with guidelines
Interacting with Computers 12 2 81ndash118Wand Y amp Weber R (1995) On the deep structure of information systems Information Systems
Journal 5 203ndash223Wiley D A (2001) The instructional use of learning objects Bloomington IN Association for
Educational Communications and Technology
Semantic learning designs 339
copy 2006 The Author Journal compilation copy 2006 British Educational Communications and Technology Agency
The concept of oc_Role accounts not only for participant roles but also for otherkinds of lsquocomponentsrsquo This is why two specialisations of the concept ofoc_ActorSlot labelled Staff and Learner are required to differentiate partici-pants from other elements As is obvious from the earlier discussion the built-in rep-resentation of activities in OpenCyc is powerful enough for the definition of learningdesign activities
The final part of the ontological specification described so far is that of the mappingfrom LD specifications to the actual past or present activity realisations Theoc_situationConformsToSpecification predicate can be used to trace whichactivity realizations come from a given learning design activity definition and the sameoccurs with the oc_thingSpecified for other learning design elements such asroles or environments Existing learning design execution engines as CopperCore(httpcoppercoreorg) actually map LD specifications to online collaboration so thatthe translation of the ontological representations of learning designs to ontology defi-nitions of activities is straightforward following similar representation
Describing objectives outcomes and preconditionsAccording to IMS learning design learning objectives and prerequisites lsquocan bedescribed using the IMS Reusable Definition of Competency or Educational Objective(RDCEO) format but can also refer to simple resources (eg a text) with a descriptionof the learning objectiversquo As the latter option is not useful from the viewpoint of com-putational semantics we will adhere to the former As commented earlier the compe-tency view is not exclusive to other kinds of objectives related to other ontologicalstandpoints (Sicilia amp Lytras 2005) and other conceptions of change or improvementmay include social ties cognitive structures or the combination of the knowledge of agroup In fact competencies adhere mostly to behaviouristic frameworks that focus onobservable external behaviour (Tennant 1991) rather than on cognitive representa-tions Nonetheless the further discussion is equally applicable to other kinds of objec-tives so that we will adhere to the competency view
oc_Competence in OpenCyc is defined as a quantity that lsquofor describing the level ofskill with which an agent performs some taskrsquo This does not remove the lack of preci-sion that is yet present in RDCEO definitions as described by Sicilia (2005) As analternative Figure 1a provides a simplified account of the model described by Sicilia asrelated to OpenCyc terms in which competence is integrated as one of the elements of thebroader notion of competency according to the model of Rummel (Rothwell amp Kazanas1992) which considers work situations attributes of the individual that are applied insuch situations and measurement with respect to some standard of performance Themost relevant distinction is again the separation of actual competency representationsand competency lsquodefinitionsrsquo (ie specifications) intended to provide a type structure tocompetencies to be used in objective or precondition specification In this case thegeneric oc-thingSpecified predicate can be used for the linking of both levels ofrepresentation
340 British Journal of Educational Technology Vol 37 No 3 2006
copy 2006 The Author Journal compilation copy 2006 British Educational Communications and Technology Agency
Objectives and preconditions can then be expressed as formulas on the acquisition ofcompetencies or their components (competency elements) Descriptions specific toOpenCyc may use oc_CycLSentence-Assertible but simpler enumeration ofcompetency components could be used instead
Describing learning design rationales theories hypotheses guidelines and rulesThe concept of lsquodesignrsquo for our purposes is properly captured in the 5a definitionprovided in the Merriam Webster Online Dictionary (wwwm-wcom) lsquoan underlyingscheme that governs functioning developing or unfoldingrsquo In consequence learningdesigns refer to artefacts and not to the process of devising the educational solutionThis is equivalent to the concept oc_Design subsumed by the notion ofoc_Specification which is described as lsquoan abstract work that constitutes adescription of the properties of a Situation or a SomethingExisting and sometimeseven entire collections of such thingsrsquo In consequence the term LearningDesignwould simply delineate a subset of specifications that are oriented to describe howlearning experiences with concrete objectives should be Trivially oc_Designerdescribes the authors of the learning designs and oc_Designing represent theoc_HumanActivity that leads to the production of learning designs Designing as aspecial kind of activity is carried out (oc_doneBy) by a designer which provides a wayto attach metainformation about the rationale of the design
From an ontological perspective the departing problem in modelling learning designrationales is that their assumptions and hypotheses of different theoretical standpointsneed not be compatible or need not consider the same aspects of reality (Conole et al2004 Packer amp Goicoechea 2000) Furthermore different pedagogical approaches areknown to result in different mappings to activities (Conole et al 2004) This wouldeither require the provision of separate ontologies or the use of a representationalmechanism that allows such kind of potential inconsistency or divergence The conceptof microtheory in OpenCyc provides such representational mechanism intended to orga-nise assertions that depend on lsquoshared set of assumptions on which the truth of theassertions dependsrsquo Definitions inside the same microtheory need to be consistent butthis is not required across microtheories The descriptions provided in the earlier sectioncan be used as a base microtheory called LearningActivitiesMt for others that are spe-cific to concrete theoretical frameworks In this section we will describe elements thatare common to the description of theoretical frameworks the distinction among micro-theories will not be discussed
The intellectual processes of learning design in themselves will be represented asoc_Designing activities carried out by oc_Designers An additional consider-ation is that the information about the rationale of the design would be better attachedto representations of each of the design processes (ie to oc_Designing instances)than to the final design This enables the registering of different assumptions for differ-ent contributions to the same artefact eg different revisions from the same of differentdesigners
Semantic learning designs 341
copy 2006 The Author Journal compilation copy 2006 British Educational Communications and Technology Agency
Several designers may participate in the same designing activity and formal frameworksfor collaborative learning resource creation (Dodero Aedo amp Diacuteaz 2002) can be inte-grated by clearly specifying which class of things can be the outcome of a designingprocess For the sake of flexibility the concept of LearningDesignPart is introduced as anall-encompassing category that includes each of the elements described in the previoussection This includes not only activities and roles but also objectives conditions andthe like Furthermore the granularity of the descriptions allows for different levels fromconsidering a unique designing activity for the whole package to recording each ele-ment of the learning design comprehensively Because oc_Actions are subsumed bythe concept of oc_Event considered as a lsquodynamic situation in which the state of theworld changesrsquo each edition of a metadata record or any other form of description maybe subject to be described The concept is flexible enough to be able to record fine-grained detailed decisions This of course may include different versions of the sameartefacts but this is an orthogonal aspect of configuration that is not relevant to ourpresent discussion
Figure 2 depicts the main elements of the loci to which rationales assumptions andguidelines will be attached
A LearningDesigner can be defined as the class of humans that performactivities of learning resource design The consideration of learning design as aunique human activity could be considered as restrictive of intelligent softwarefor the task Simply changing the oc_Designer class to a subconcept ofoc_IntelligentAgent could remove this but this entails a change in Open-Cyc knowledge base The concept oc_Designing is a subclass of the conceptoc_ThinkingAboutTheStateOfTheWorld which represents lsquothinking pro-cessesrsquo which include mental events such as planning evaluating or inventingThus the concept of design described so far should be understood in terms of the
Figure 2 Main elements of the ontological schema for the loci of design rationales
oc_performedBy modifies
342 British Journal of Educational Technology Vol 37 No 3 2006
copy 2006 The Author Journal compilation copy 2006 British Educational Communications and Technology Agency
concept of lsquoexpanded rationalityrsquo as described by Hatchuel (2002) integrating cre-ativity and unexpected expansions of the original requirements This precludes onto-logical definitions in which the problem space is completely bounded a priori Inconsequence a degree of openness is necessary to integrate different kinds of detailin description from fully described ones to others with shallower semantics eg someproviding only references to generic assumptions
Theoretical frameworksWe will make the distinction that guidelines (as further defined) are the propositions thatdirectly predicate about the validity or appropriateness of learning design parts whileother theoretical propositions about learning that are not directly saying anythingabout the design of digital learning resources are considered as assumptions followingthe general sense provided in the Merriam-Webster dictionary lsquoa fact or statement (asa proposition axiom postulate or notion) taken for grantedrsquo
The fact that some assumption hypotheses or proposed law about learning is sustainedby a broader empirical base or is better sustained by rational means is out of the scopeof this work because here we focus on design practice (even though it is of an obviousinterest for researchers) In principle even unorthodox positions may be sustainedbecause the mechanism of microtheories and their inclusion mechanism allows forgreat flexibility in modelling them
Table 2 provides some examples of theoretical assumptions and related guidelines forlsquoextremersquo archetypical standpoints on learning These are not intended to reflect actualcommon positions but they are provided as illustrative examples of radical positionsthat could be modelled with our ontological schema
The examples provided in Table 2 go a step further than toolkits in Conole et al (2004)in making the design rationale explicit Nonetheless in many cases lsquocommon sensersquo isapplied to design and a shallower approach to specification could be used instead egproviding simply links to the theoretical standpoints used It should be noted also thatthe resulting guidelines vary greatly in their level of concreteness and some of themmay be considered as mandatory from some standpoints while others are mere generalrepresentations This would lead to different computational strategies
Guidelines and rulesThe WordNet (httpwordnetprincetonedu) thesaurus defines guidelines as lsquoguidancerelative to setting standards or determining a course of actionrsquo or lsquoa rule or principlethat provides guidance to appropriate behaviorrsquo In our case guidelines are statementsthat serve as criteria for decision in designing learning Given that e-learning designis still far from being a mature discipline guidelinesmdashas instances of recordedknowledgemdashcan be characterised as oc_NormativeSpecifications becausethey describe how learning resource design activities and their outcomes lsquoshould bersquoin a propositional form Propositions lsquohave some truth value in some context or
Semantic learning designs 343
copy 2006 The Author Journal compilation copy 2006 British Educational Communications and Technology Agency
ldquoworldrdquo rsquo Furthermore propositions lsquoare assumed to be representable (at least in prin-ciple) by a sentence in some formal or natural languagersquo This raises the need for alanguage that expresses guidelines in a general sense There are two main options forsuch representation be them formal logical sentences or reified guidelines
Formal logical sentence representation entails the formalisation through sentences ofthe constraints of the designs For example if lsquoevery learning resource must be activersquoaccording to some theoretical standpoint (eg one that fosters reflective learning) thissentence may be represented as an axiom and put into the appropriate separate micro-theory as the following CycL fragment
($forAll X ($implies
($isa X $LearningObject)
($resourceType X $Active))
Then resources that are not active are simply excluded from the notion itself This is astrong position in the sense that precludes some views on the learning design process
Table 2 Some examples of theoretical assumptions and guidelines
Archetypicalstandpoint Assumptions Guidelines
Instructivist(Behaviouristic)
bull Learning occurs through association and reinforcement
bull The results of learning can be appreciated through observable outcomes
(GB1) Every instance of oc-Learning entails theexposure to some LearningObjectthat produces the association (GB2) Learning Activitysequences should provide assessmentactivities that confirm theassociations that should have beenestablished so far
Constructivist bull Learning proceeds by building mental structures
(GC1) Agentrsquos mental structuresshould be represented(GC2) These structures should bepre-assessed in order to becomparable with the resulting ones(GC3) Design Activities arerecommended(GC4) Contrast Activities arerecommended
Sociocultural bull Learning occurs through social participatio
(GSC1) Every instance ofoc-Learning entails theinteraction of severaloc-IntelligentAgents asoc-Actors
344 British Journal of Educational Technology Vol 37 No 3 2006
copy 2006 The Author Journal compilation copy 2006 British Educational Communications and Technology Agency
Another typical example may be that of a sociocultural guideline that preclude Methodsthat do not entail multirole activities that provide access to social interaction ServicesThis approach to guidelines provides the best framework for automation and wouldenable the creation of microtheories that represent clearly formulated views onlearning Nonetheless there are two major problems for the use of this option as theunique mechanism for encoding design rationales
1 There is no clear-cut distinction between models and theories of learning becausethey come from a history of schools of thought with interrelations (Palmer 2001)This would lead to the need for the representation of dozens of variants reflectingpositions that only diverge in some aspects
2 Many guidelines cannot be represented as logical sentences because theyrepresent lsquorecommendationsrsquo or vague requirements eg see (GC3) or (GC4)(Table 2)
These issues suggest that a combination of formal sentences with reified guidelinesprovide the best option for flexibility The representation of guidelines has beenstudied in the field of HumanndashComputer-Interaction (HCI) in the last decade(Vanderdonckt 1999) and the main philosophy of such representations remainsvalid in the learning design domain The major challenge for such representations isthat of providing computational semantics that could be used for the implementationof software support but this is difficult to attain in all the cases Thus the concept oflearning design guideline will provide two kinds of representations general andschema-specific
The general use of instances of oc_NormativeSpecification allows the model-ling of lsquoconceptual works that describe how something should bersquo This may includerepresentation of known normative representations or procedures For example onemay refer to a concrete learning design model as that derived from Bannister (1981)which states that creativity cycles follow an initial phase of brainstorming (idea cap-ture) followed by a phase in which key issues are focused into an operation strategy(idea development) and a third phase in which project control takes place through alsquorecursive cyclersquo (operational management) The reference could be made by pointing tothe academic reference describing it as a conceptual work or it could be described asan Activity-Spec modelled explicitly inside the ontology As another examplesome detailed guidelines could be modelled by the normative specification sub-category of oc-Instructions defined as lsquoa sequence of tasks to be performed byan agentrsquo
Schema-specific guidelines restrict the freedom to represent criteria as conceptual workinstances to a concrete and commonly agreed ontology that determine what can be saidabout how learning designs should be according to a theoretical standpoint or frame-work These build on an activityndashrole ontology as the one described in the previoussection and add a prescription on the kind of lsquoshould-bersquo predications that are allow-able Table 3 describes examples of a minimal language that could be used for thatpurpose
Semantic learning designs 345
copy 2006 The Author Journal compilation copy 2006 British Educational Communications and Technology Agency
In the examples CAT represents a category of individual expressed through a conceptor a logical formula and the sentence is a synthetic way of expressing some recom-mended or mandatory constraint over such category Other languages could be engi-neered in further research but the one described can be directly expressed in OpenCycand combined with software modules that check or enforce them in some way Notethat this representation does not use direct OpenCyc sentences so that reification is usedinstead of direct logical sentences It is not the aim of this paper to determine the mostpowerful and flexible representation language but only set a foundation for furtherspecification efforts
Analysis and comparison of two theoretical semantic frameworksAs an assessment of the representational capabilities of the scheme just described acase study of representation is sketched in this section We will use the learning designtoolkit described by Conole et al (2004) as the underlying terminology even thoughother frameworks could be used instead
Two rather diverging positions for design will be used as an illustration Table 4 sum-marises the two design scenarios
Table 3 Possible language for use in learning designs
Sentence type Potential uses Example
RecommendationshouldBe (CATSENT)
Recommend kinds ofactivities
Design activities are recommended (GC3)shouldBe(Method subEvent (CATDesignActivity))
ObligationmustBe (CATSENT)
Some activity typesmust follow others
Assessments must follow each expositionto contents (GD1) mustBe(and(ExpositiveActivityIndividualActivity)startAfterEndingOf(CAT
AssessmentActivity)
Activities must havesome characteristic
Every activity must entail interactionmustBe(LDActivity(usedIn(CAT
InteractiveService)))Every activity must provide acontent-oriented learning object forindividual use (similar to GB1)mustBe(LDActivity(and (usedIn(CAT
LearningObject)(learnerRoleNumber(1))))
Some structure shouldbe enforced in methods
Methods should have a sequential actstructuremustBe(Methodnot ((subEvent(CAT Play))))
CAT = category GC3 = Constructivist Guideline 3 (see Table 2) GD1 = Guideline 1
346 British Journal of Educational Technology Vol 37 No 3 2006
copy 2006 The Author Journal compilation copy 2006 British Educational Communications and Technology Agency
The following subsections sketch the main outcomes of an example learning designprocess for both scenarios assuming that the process described by Conole et al (2004)is used in both situations
Differences in modelling the design processThe forces driving the design process in both scenarios are fairly diverging The designof Scenario A will typically be constrained by a clear target specified in terms of anaggregated competency level while in Scenario B there are no concrete competenciesto be acquired and the main objective of the design activity would be twofold On theone hand the activities are targeted at promoting interchange of ideas in an attemptto stimulate casual convergence of interest which is the main value of the culture ofthe special interest group On the other hand the activities have the generic objectiveof strengthening social ties which is an important element of cohesion for the under-taking of joint projects
Table 5 describes the main inputs and possible activities and their objectives at a highlevel
The changes intended to be achieved in each of the scenarios are stated differently witha formal definition in (A) and a shallow cognitive and social representation in the case
Table 4 Learning design scenarios
Scenario DescriptionClassification according to
Conole et al (2004) toolkit
(A) Systemic-Behaviourist A software development company attempts to designlearning units about newtechnologies or softwareproducts for its developerstaff The design is drivenby a cycle of lsquocompetencygap analysisrsquo that modelsorganisational learning in terms of the aggregated competencies of the employeesmdashsimilar to theone described by Sicilia(2005)
Focuses on the individual where a body of informationforms the basis and rawmaterial for learning Learning is mainly considered as nonreflective ie based on skill learning and memorisation
(B) Sociocultural A special interest group ona concrete research topicaims at developing the knowledge of its membersby sharing experiencesand discussing on themesof common interest
Focuses on the social whereexperience is the main source of learning Conscious reflection on experience is transformed into learning
Semantic learning designs 347
copy 2006 The Author Journal compilation copy 2006 British Educational Communications and Technology Agency
of (B) In addition case (B) requires a model of social network that is largely irrelevantin case (A)
Differences in recording activity rationalesThe approach for the recording of design rationales also differs because of the differentcontexts Table 6 summarises the main elements for each of the scenarios
Conclusions and directions for researchCurrent technology and practice related to the description of reusable digital learningresources has resulted in techniques that allow the description of learning activities of
Table 5 Inputs activities and objectives of two learning design scenarios
Scenario Design inputs Activities and their objectives
(A) Systemic-Behaviourist A formal competency recordfor employees and competencyrequirements expressed in terms of amounts of required competencies (assuming a shared measurement scale) Such requirements are actually derived from strategy or from the contingent requirements of upcoming projects
Competency instances are used to measure the levelof Competency-Spec for the required competencies (or their constituents) A formal match considering the agenda of employees is carried out for determining the target learners LearningObjects forindividual use are selected asthe main elements of theactivities and the criteria forselection are again a match ofthe objectives stated inmetadata (as stated incategory 9 of Learning Object MetadatamdashClassification) and thecompetencies required
(B) Sociocultural The expertise of the special interest group members as stated in the concepts addressed by their publicationsprojects or other activities
Informal group activities are the main approach applicablesince there is not a predefinedgroup structure but a cultureof sharing and socialisation The results of the sharing of ideas can be expressed in terms of oc-knowsAboutpredicates and social ties will be modelled through positive valence FOAF-knows(httpxmlnscomfoaf01)predicates with an added numerical value representingstrength
348 British Journal of Educational Technology Vol 37 No 3 2006
copy 2006 The Author Journal compilation copy 2006 British Educational Communications and Technology Agency
a complex kind including their intended participants educational objectives and exe-cution constraints The representation and recording of the assumptions and guidelinesthat were considered in the creation of such resources would extend current techniquesto capture the rationale of the design itself This in turn will eventually result in theavailability of a large base of empirical data available in which theoretical assumptionsor rules are explicitly connected to the resulting designs as these could be linked withthe actual results and trace of the learning processes that used them The resulting bodyof evidence would represent a critical asset in the development of the theory of learningmediated through computers
This paper has described the first attempt to extend existing learning technology modelswith a schema explicitly targeted at describing learning design rationales The OpenCycknowledge base has been used as the semantic reference from which the ontologicalschema has been specified Concretely an activity-based schema based on IMS LD hasbeen described and theoretical assumptions and codified guidelines have been intro-duced The flexibility of the representation schema has been analysed through therepresentation of two stereotyped theoretical positions
Two main research directions should continue the work described here On the onehand further ontological engineering effort is required to come up with a more com-prehensive and rich schema for learning design rationales On the other hand tool
Table 6 Main elements of the two learning design scenarios
Scenario Approach Examples
(A) Systemic-Behaviourist Formal logical sentencesare used to guarantee that every design followsan organisationalobjective
It is required that the objective of every Method is attached to a Competency-Spec instance as objective
Formal assessment is a requirement foraccountability purposes
Guideline (GD1) is attached to everyDesigningLearningResource instance that produces an Act
(B) Sociocultural The interaction in informal settings can beused to derive social models (Sicilia amp Garciacutea2004)
The interactions in discussion canbe used to measure the degree of lsquoproximityrsquo between members Therationale of the activities can bejustified in terms of SocialTies inside SocialNetworks Further the focus of each groupactivity should be clearly stated anda combination of the increase in collective knowledge and the strengthening of social relations can be used for assessment
Semantic learning designs 349
copy 2006 The Author Journal compilation copy 2006 British Educational Communications and Technology Agency
implementation and record gathering efforts would be required to attain the ultimateobjective of the ontological definitions described so far With such effort an unprece-dented research material could be available in some years providing improved oppor-tunities for scientific inquiry on design as an activity regarding pedagogy
Other potential directions for research may be the description of interpretations of thedesign intentions (for which no rationale was explicitly recorded) the mining or detec-tion of patterns as related to rationale description or the detailed use of theoreticalrationales in agents that support the learning process which represent a category ofrelevant learning support software (Holmes in press)
ReferencesAllert H (2004) Coherent social systems for learningmdashan approach for contextualized and
community-centred metadata Journal of Interactive Media in Education 2004 1Baader F Calvanese D McGuinness D Nardi D amp Patel-Schneider P (Eds) (2003) The
description logic handbook Theory implementation and applications Cambridge CambridgeUniversity Press
Bannister D (1981) Personal construct theory and research method In P Reason amp J Rowan(Eds) Human inquiry a sourcebook of new paradigm research (pp 191ndash200) Chichester UKJohn Wiley and Sons
Berners-Lee T Hendler J amp Lassila O (2001) The semantic web Scientific American 284 534ndash43
Brown A amp Voltz B (2005) Elements of effective e-learning design International Review ofResearch in Open and Distance Learning 6 1
Buzza D Bean D Harrigan K amp Carey T (2005) Learning design repositories adaptinglearning design specifications for shared instructional knowledge Canadian Journal of Learningand Technology 30 3
Conole G Dyke M Oliver M amp Seale J (2004) Mapping pedagogy and tools for effectivelearning design Computers amp Education 43 1ndash2 17ndash33
Dodero J M Aedo I amp Diacuteaz P (2002) Participative knowledge production of learning objectsfor e-books The Electronic Library 20 4 296ndash305
Friesen N (2005) Interoperability and learning objects an overview of e-learning standardiza-tion Interdisciplinary Journal of Knowledge and Learning Objects 1 1 23ndash31
Gruber T (1995) Towards principles for the design of ontologies used for knowledge sharingInternational Journal of HumanndashComputer Studies 43 56 907ndash928
Hatchuel A (2002) Towards design theory and expandable rationality the unfinished programof Herbert Simon Journal of Management and Governance 5 3ndash4 260ndash271
Holmes J (in press) Designing agents to support learning by explaining Computers and EducationIEEE Learning Technology Standards Committee (2002) Learning object metadata (LOM) Final
Draft Standard IEEE 1484121ndash2002IMS (2003) IMS Learning Design Information Model (Version 10) Final Specification 20
January 2003Kaptelinin V Kuutti K amp Bannon L (1995) Activity theory basic concepts and applications
In Blumenthalet al (Eds) Humanndashcomputer interaction Lecture Notes in Computer Science 1015189ndash201
Koper R (2004) Use of the semantic web to solve some basic problems in education increaseflexible distributed lifelong learning decrease teacherrsquos workload Journal of Interactive Mediain Education 2004 6
Lenat D (1995) CYC a large-scale investment in knowledge infrastructure Communications ofthe ACM 38 11 33ndash38
350 British Journal of Educational Technology Vol 37 No 3 2006
copy 2006 The Author Journal compilation copy 2006 British Educational Communications and Technology Agency
McGreal R (2004) Learning objects a practical definition International Journal of InstructionalTechnology and Distance Learning 1 9 21ndash32
Packer M amp Goicoechea J (2000) Sociocultural and constructivist theories of learningontology not just epistemology Educational Psychologist 35 4 227ndash241
Palmer J (2001) Fifty major thinkers on education London and New York RoutledgePolsani P R (2003) Use and abuse of reusable learning objects Journal of Digital Information 3
4 Rothwell W amp Kazanas H (1992) Mastering the instructional design process San Francisco CA
Jossey-BassSaacutenchez-Alonso S amp Sicilia M A (2004) Relationships and commitments in learning object
metadata In Proceedings of the 5th International Conference on Information Technology BasedHigher Education and Training ITHET 2004 Istanbul Turkey
Sicilia M A (2005) Ontology-Based Competency Management Infrastructures for theKnowledge-intensive Learning Organization In Lytras amp Naeve (Eds) Intelligent Learning Infra-structures in Knowledge Intensive Organizations A Semantic Web perspective IDEA USA pp 302ndash324
Sicilia M A amp Garciacutea E (2004) Fuzzy group models for adaptation in cooperative informationretrieval contexts Lecture Notes in Computer Science 2932 324ndash334
Sicilia M A amp Garciacutea E (2005) On the convergence of formal ontologies and standardizede-learning Journal of Distance Education Technologies 3 2 13ndash29
Sicilia M A amp Lytras M (2005) On the representation of change according to different ontol-ogies of learning International Journal of Learning and Change 1 1 66ndash79
Sicilia M A Lytras M Rodriacuteguez E amp Garciacutea E (in press) Integrating descriptions of knowl-edge management learning activities into large ontological structures a case study Data andKnowledge Engineering
Tennant M (1991) Expertise as a dimension of adult development New Education 13 1 49ndash55Vanderdonckt J (1999) Development milestones towards a tool for working with guidelines
Interacting with Computers 12 2 81ndash118Wand Y amp Weber R (1995) On the deep structure of information systems Information Systems
Journal 5 203ndash223Wiley D A (2001) The instructional use of learning objects Bloomington IN Association for
Educational Communications and Technology
340 British Journal of Educational Technology Vol 37 No 3 2006
copy 2006 The Author Journal compilation copy 2006 British Educational Communications and Technology Agency
Objectives and preconditions can then be expressed as formulas on the acquisition ofcompetencies or their components (competency elements) Descriptions specific toOpenCyc may use oc_CycLSentence-Assertible but simpler enumeration ofcompetency components could be used instead
Describing learning design rationales theories hypotheses guidelines and rulesThe concept of lsquodesignrsquo for our purposes is properly captured in the 5a definitionprovided in the Merriam Webster Online Dictionary (wwwm-wcom) lsquoan underlyingscheme that governs functioning developing or unfoldingrsquo In consequence learningdesigns refer to artefacts and not to the process of devising the educational solutionThis is equivalent to the concept oc_Design subsumed by the notion ofoc_Specification which is described as lsquoan abstract work that constitutes adescription of the properties of a Situation or a SomethingExisting and sometimeseven entire collections of such thingsrsquo In consequence the term LearningDesignwould simply delineate a subset of specifications that are oriented to describe howlearning experiences with concrete objectives should be Trivially oc_Designerdescribes the authors of the learning designs and oc_Designing represent theoc_HumanActivity that leads to the production of learning designs Designing as aspecial kind of activity is carried out (oc_doneBy) by a designer which provides a wayto attach metainformation about the rationale of the design
From an ontological perspective the departing problem in modelling learning designrationales is that their assumptions and hypotheses of different theoretical standpointsneed not be compatible or need not consider the same aspects of reality (Conole et al2004 Packer amp Goicoechea 2000) Furthermore different pedagogical approaches areknown to result in different mappings to activities (Conole et al 2004) This wouldeither require the provision of separate ontologies or the use of a representationalmechanism that allows such kind of potential inconsistency or divergence The conceptof microtheory in OpenCyc provides such representational mechanism intended to orga-nise assertions that depend on lsquoshared set of assumptions on which the truth of theassertions dependsrsquo Definitions inside the same microtheory need to be consistent butthis is not required across microtheories The descriptions provided in the earlier sectioncan be used as a base microtheory called LearningActivitiesMt for others that are spe-cific to concrete theoretical frameworks In this section we will describe elements thatare common to the description of theoretical frameworks the distinction among micro-theories will not be discussed
The intellectual processes of learning design in themselves will be represented asoc_Designing activities carried out by oc_Designers An additional consider-ation is that the information about the rationale of the design would be better attachedto representations of each of the design processes (ie to oc_Designing instances)than to the final design This enables the registering of different assumptions for differ-ent contributions to the same artefact eg different revisions from the same of differentdesigners
Semantic learning designs 341
copy 2006 The Author Journal compilation copy 2006 British Educational Communications and Technology Agency
Several designers may participate in the same designing activity and formal frameworksfor collaborative learning resource creation (Dodero Aedo amp Diacuteaz 2002) can be inte-grated by clearly specifying which class of things can be the outcome of a designingprocess For the sake of flexibility the concept of LearningDesignPart is introduced as anall-encompassing category that includes each of the elements described in the previoussection This includes not only activities and roles but also objectives conditions andthe like Furthermore the granularity of the descriptions allows for different levels fromconsidering a unique designing activity for the whole package to recording each ele-ment of the learning design comprehensively Because oc_Actions are subsumed bythe concept of oc_Event considered as a lsquodynamic situation in which the state of theworld changesrsquo each edition of a metadata record or any other form of description maybe subject to be described The concept is flexible enough to be able to record fine-grained detailed decisions This of course may include different versions of the sameartefacts but this is an orthogonal aspect of configuration that is not relevant to ourpresent discussion
Figure 2 depicts the main elements of the loci to which rationales assumptions andguidelines will be attached
A LearningDesigner can be defined as the class of humans that performactivities of learning resource design The consideration of learning design as aunique human activity could be considered as restrictive of intelligent softwarefor the task Simply changing the oc_Designer class to a subconcept ofoc_IntelligentAgent could remove this but this entails a change in Open-Cyc knowledge base The concept oc_Designing is a subclass of the conceptoc_ThinkingAboutTheStateOfTheWorld which represents lsquothinking pro-cessesrsquo which include mental events such as planning evaluating or inventingThus the concept of design described so far should be understood in terms of the
Figure 2 Main elements of the ontological schema for the loci of design rationales
oc_performedBy modifies
342 British Journal of Educational Technology Vol 37 No 3 2006
copy 2006 The Author Journal compilation copy 2006 British Educational Communications and Technology Agency
concept of lsquoexpanded rationalityrsquo as described by Hatchuel (2002) integrating cre-ativity and unexpected expansions of the original requirements This precludes onto-logical definitions in which the problem space is completely bounded a priori Inconsequence a degree of openness is necessary to integrate different kinds of detailin description from fully described ones to others with shallower semantics eg someproviding only references to generic assumptions
Theoretical frameworksWe will make the distinction that guidelines (as further defined) are the propositions thatdirectly predicate about the validity or appropriateness of learning design parts whileother theoretical propositions about learning that are not directly saying anythingabout the design of digital learning resources are considered as assumptions followingthe general sense provided in the Merriam-Webster dictionary lsquoa fact or statement (asa proposition axiom postulate or notion) taken for grantedrsquo
The fact that some assumption hypotheses or proposed law about learning is sustainedby a broader empirical base or is better sustained by rational means is out of the scopeof this work because here we focus on design practice (even though it is of an obviousinterest for researchers) In principle even unorthodox positions may be sustainedbecause the mechanism of microtheories and their inclusion mechanism allows forgreat flexibility in modelling them
Table 2 provides some examples of theoretical assumptions and related guidelines forlsquoextremersquo archetypical standpoints on learning These are not intended to reflect actualcommon positions but they are provided as illustrative examples of radical positionsthat could be modelled with our ontological schema
The examples provided in Table 2 go a step further than toolkits in Conole et al (2004)in making the design rationale explicit Nonetheless in many cases lsquocommon sensersquo isapplied to design and a shallower approach to specification could be used instead egproviding simply links to the theoretical standpoints used It should be noted also thatthe resulting guidelines vary greatly in their level of concreteness and some of themmay be considered as mandatory from some standpoints while others are mere generalrepresentations This would lead to different computational strategies
Guidelines and rulesThe WordNet (httpwordnetprincetonedu) thesaurus defines guidelines as lsquoguidancerelative to setting standards or determining a course of actionrsquo or lsquoa rule or principlethat provides guidance to appropriate behaviorrsquo In our case guidelines are statementsthat serve as criteria for decision in designing learning Given that e-learning designis still far from being a mature discipline guidelinesmdashas instances of recordedknowledgemdashcan be characterised as oc_NormativeSpecifications becausethey describe how learning resource design activities and their outcomes lsquoshould bersquoin a propositional form Propositions lsquohave some truth value in some context or
Semantic learning designs 343
copy 2006 The Author Journal compilation copy 2006 British Educational Communications and Technology Agency
ldquoworldrdquo rsquo Furthermore propositions lsquoare assumed to be representable (at least in prin-ciple) by a sentence in some formal or natural languagersquo This raises the need for alanguage that expresses guidelines in a general sense There are two main options forsuch representation be them formal logical sentences or reified guidelines
Formal logical sentence representation entails the formalisation through sentences ofthe constraints of the designs For example if lsquoevery learning resource must be activersquoaccording to some theoretical standpoint (eg one that fosters reflective learning) thissentence may be represented as an axiom and put into the appropriate separate micro-theory as the following CycL fragment
($forAll X ($implies
($isa X $LearningObject)
($resourceType X $Active))
Then resources that are not active are simply excluded from the notion itself This is astrong position in the sense that precludes some views on the learning design process
Table 2 Some examples of theoretical assumptions and guidelines
Archetypicalstandpoint Assumptions Guidelines
Instructivist(Behaviouristic)
bull Learning occurs through association and reinforcement
bull The results of learning can be appreciated through observable outcomes
(GB1) Every instance of oc-Learning entails theexposure to some LearningObjectthat produces the association (GB2) Learning Activitysequences should provide assessmentactivities that confirm theassociations that should have beenestablished so far
Constructivist bull Learning proceeds by building mental structures
(GC1) Agentrsquos mental structuresshould be represented(GC2) These structures should bepre-assessed in order to becomparable with the resulting ones(GC3) Design Activities arerecommended(GC4) Contrast Activities arerecommended
Sociocultural bull Learning occurs through social participatio
(GSC1) Every instance ofoc-Learning entails theinteraction of severaloc-IntelligentAgents asoc-Actors
344 British Journal of Educational Technology Vol 37 No 3 2006
copy 2006 The Author Journal compilation copy 2006 British Educational Communications and Technology Agency
Another typical example may be that of a sociocultural guideline that preclude Methodsthat do not entail multirole activities that provide access to social interaction ServicesThis approach to guidelines provides the best framework for automation and wouldenable the creation of microtheories that represent clearly formulated views onlearning Nonetheless there are two major problems for the use of this option as theunique mechanism for encoding design rationales
1 There is no clear-cut distinction between models and theories of learning becausethey come from a history of schools of thought with interrelations (Palmer 2001)This would lead to the need for the representation of dozens of variants reflectingpositions that only diverge in some aspects
2 Many guidelines cannot be represented as logical sentences because theyrepresent lsquorecommendationsrsquo or vague requirements eg see (GC3) or (GC4)(Table 2)
These issues suggest that a combination of formal sentences with reified guidelinesprovide the best option for flexibility The representation of guidelines has beenstudied in the field of HumanndashComputer-Interaction (HCI) in the last decade(Vanderdonckt 1999) and the main philosophy of such representations remainsvalid in the learning design domain The major challenge for such representations isthat of providing computational semantics that could be used for the implementationof software support but this is difficult to attain in all the cases Thus the concept oflearning design guideline will provide two kinds of representations general andschema-specific
The general use of instances of oc_NormativeSpecification allows the model-ling of lsquoconceptual works that describe how something should bersquo This may includerepresentation of known normative representations or procedures For example onemay refer to a concrete learning design model as that derived from Bannister (1981)which states that creativity cycles follow an initial phase of brainstorming (idea cap-ture) followed by a phase in which key issues are focused into an operation strategy(idea development) and a third phase in which project control takes place through alsquorecursive cyclersquo (operational management) The reference could be made by pointing tothe academic reference describing it as a conceptual work or it could be described asan Activity-Spec modelled explicitly inside the ontology As another examplesome detailed guidelines could be modelled by the normative specification sub-category of oc-Instructions defined as lsquoa sequence of tasks to be performed byan agentrsquo
Schema-specific guidelines restrict the freedom to represent criteria as conceptual workinstances to a concrete and commonly agreed ontology that determine what can be saidabout how learning designs should be according to a theoretical standpoint or frame-work These build on an activityndashrole ontology as the one described in the previoussection and add a prescription on the kind of lsquoshould-bersquo predications that are allow-able Table 3 describes examples of a minimal language that could be used for thatpurpose
Semantic learning designs 345
copy 2006 The Author Journal compilation copy 2006 British Educational Communications and Technology Agency
In the examples CAT represents a category of individual expressed through a conceptor a logical formula and the sentence is a synthetic way of expressing some recom-mended or mandatory constraint over such category Other languages could be engi-neered in further research but the one described can be directly expressed in OpenCycand combined with software modules that check or enforce them in some way Notethat this representation does not use direct OpenCyc sentences so that reification is usedinstead of direct logical sentences It is not the aim of this paper to determine the mostpowerful and flexible representation language but only set a foundation for furtherspecification efforts
Analysis and comparison of two theoretical semantic frameworksAs an assessment of the representational capabilities of the scheme just described acase study of representation is sketched in this section We will use the learning designtoolkit described by Conole et al (2004) as the underlying terminology even thoughother frameworks could be used instead
Two rather diverging positions for design will be used as an illustration Table 4 sum-marises the two design scenarios
Table 3 Possible language for use in learning designs
Sentence type Potential uses Example
RecommendationshouldBe (CATSENT)
Recommend kinds ofactivities
Design activities are recommended (GC3)shouldBe(Method subEvent (CATDesignActivity))
ObligationmustBe (CATSENT)
Some activity typesmust follow others
Assessments must follow each expositionto contents (GD1) mustBe(and(ExpositiveActivityIndividualActivity)startAfterEndingOf(CAT
AssessmentActivity)
Activities must havesome characteristic
Every activity must entail interactionmustBe(LDActivity(usedIn(CAT
InteractiveService)))Every activity must provide acontent-oriented learning object forindividual use (similar to GB1)mustBe(LDActivity(and (usedIn(CAT
LearningObject)(learnerRoleNumber(1))))
Some structure shouldbe enforced in methods
Methods should have a sequential actstructuremustBe(Methodnot ((subEvent(CAT Play))))
CAT = category GC3 = Constructivist Guideline 3 (see Table 2) GD1 = Guideline 1
346 British Journal of Educational Technology Vol 37 No 3 2006
copy 2006 The Author Journal compilation copy 2006 British Educational Communications and Technology Agency
The following subsections sketch the main outcomes of an example learning designprocess for both scenarios assuming that the process described by Conole et al (2004)is used in both situations
Differences in modelling the design processThe forces driving the design process in both scenarios are fairly diverging The designof Scenario A will typically be constrained by a clear target specified in terms of anaggregated competency level while in Scenario B there are no concrete competenciesto be acquired and the main objective of the design activity would be twofold On theone hand the activities are targeted at promoting interchange of ideas in an attemptto stimulate casual convergence of interest which is the main value of the culture ofthe special interest group On the other hand the activities have the generic objectiveof strengthening social ties which is an important element of cohesion for the under-taking of joint projects
Table 5 describes the main inputs and possible activities and their objectives at a highlevel
The changes intended to be achieved in each of the scenarios are stated differently witha formal definition in (A) and a shallow cognitive and social representation in the case
Table 4 Learning design scenarios
Scenario DescriptionClassification according to
Conole et al (2004) toolkit
(A) Systemic-Behaviourist A software development company attempts to designlearning units about newtechnologies or softwareproducts for its developerstaff The design is drivenby a cycle of lsquocompetencygap analysisrsquo that modelsorganisational learning in terms of the aggregated competencies of the employeesmdashsimilar to theone described by Sicilia(2005)
Focuses on the individual where a body of informationforms the basis and rawmaterial for learning Learning is mainly considered as nonreflective ie based on skill learning and memorisation
(B) Sociocultural A special interest group ona concrete research topicaims at developing the knowledge of its membersby sharing experiencesand discussing on themesof common interest
Focuses on the social whereexperience is the main source of learning Conscious reflection on experience is transformed into learning
Semantic learning designs 347
copy 2006 The Author Journal compilation copy 2006 British Educational Communications and Technology Agency
of (B) In addition case (B) requires a model of social network that is largely irrelevantin case (A)
Differences in recording activity rationalesThe approach for the recording of design rationales also differs because of the differentcontexts Table 6 summarises the main elements for each of the scenarios
Conclusions and directions for researchCurrent technology and practice related to the description of reusable digital learningresources has resulted in techniques that allow the description of learning activities of
Table 5 Inputs activities and objectives of two learning design scenarios
Scenario Design inputs Activities and their objectives
(A) Systemic-Behaviourist A formal competency recordfor employees and competencyrequirements expressed in terms of amounts of required competencies (assuming a shared measurement scale) Such requirements are actually derived from strategy or from the contingent requirements of upcoming projects
Competency instances are used to measure the levelof Competency-Spec for the required competencies (or their constituents) A formal match considering the agenda of employees is carried out for determining the target learners LearningObjects forindividual use are selected asthe main elements of theactivities and the criteria forselection are again a match ofthe objectives stated inmetadata (as stated incategory 9 of Learning Object MetadatamdashClassification) and thecompetencies required
(B) Sociocultural The expertise of the special interest group members as stated in the concepts addressed by their publicationsprojects or other activities
Informal group activities are the main approach applicablesince there is not a predefinedgroup structure but a cultureof sharing and socialisation The results of the sharing of ideas can be expressed in terms of oc-knowsAboutpredicates and social ties will be modelled through positive valence FOAF-knows(httpxmlnscomfoaf01)predicates with an added numerical value representingstrength
348 British Journal of Educational Technology Vol 37 No 3 2006
copy 2006 The Author Journal compilation copy 2006 British Educational Communications and Technology Agency
a complex kind including their intended participants educational objectives and exe-cution constraints The representation and recording of the assumptions and guidelinesthat were considered in the creation of such resources would extend current techniquesto capture the rationale of the design itself This in turn will eventually result in theavailability of a large base of empirical data available in which theoretical assumptionsor rules are explicitly connected to the resulting designs as these could be linked withthe actual results and trace of the learning processes that used them The resulting bodyof evidence would represent a critical asset in the development of the theory of learningmediated through computers
This paper has described the first attempt to extend existing learning technology modelswith a schema explicitly targeted at describing learning design rationales The OpenCycknowledge base has been used as the semantic reference from which the ontologicalschema has been specified Concretely an activity-based schema based on IMS LD hasbeen described and theoretical assumptions and codified guidelines have been intro-duced The flexibility of the representation schema has been analysed through therepresentation of two stereotyped theoretical positions
Two main research directions should continue the work described here On the onehand further ontological engineering effort is required to come up with a more com-prehensive and rich schema for learning design rationales On the other hand tool
Table 6 Main elements of the two learning design scenarios
Scenario Approach Examples
(A) Systemic-Behaviourist Formal logical sentencesare used to guarantee that every design followsan organisationalobjective
It is required that the objective of every Method is attached to a Competency-Spec instance as objective
Formal assessment is a requirement foraccountability purposes
Guideline (GD1) is attached to everyDesigningLearningResource instance that produces an Act
(B) Sociocultural The interaction in informal settings can beused to derive social models (Sicilia amp Garciacutea2004)
The interactions in discussion canbe used to measure the degree of lsquoproximityrsquo between members Therationale of the activities can bejustified in terms of SocialTies inside SocialNetworks Further the focus of each groupactivity should be clearly stated anda combination of the increase in collective knowledge and the strengthening of social relations can be used for assessment
Semantic learning designs 349
copy 2006 The Author Journal compilation copy 2006 British Educational Communications and Technology Agency
implementation and record gathering efforts would be required to attain the ultimateobjective of the ontological definitions described so far With such effort an unprece-dented research material could be available in some years providing improved oppor-tunities for scientific inquiry on design as an activity regarding pedagogy
Other potential directions for research may be the description of interpretations of thedesign intentions (for which no rationale was explicitly recorded) the mining or detec-tion of patterns as related to rationale description or the detailed use of theoreticalrationales in agents that support the learning process which represent a category ofrelevant learning support software (Holmes in press)
ReferencesAllert H (2004) Coherent social systems for learningmdashan approach for contextualized and
community-centred metadata Journal of Interactive Media in Education 2004 1Baader F Calvanese D McGuinness D Nardi D amp Patel-Schneider P (Eds) (2003) The
description logic handbook Theory implementation and applications Cambridge CambridgeUniversity Press
Bannister D (1981) Personal construct theory and research method In P Reason amp J Rowan(Eds) Human inquiry a sourcebook of new paradigm research (pp 191ndash200) Chichester UKJohn Wiley and Sons
Berners-Lee T Hendler J amp Lassila O (2001) The semantic web Scientific American 284 534ndash43
Brown A amp Voltz B (2005) Elements of effective e-learning design International Review ofResearch in Open and Distance Learning 6 1
Buzza D Bean D Harrigan K amp Carey T (2005) Learning design repositories adaptinglearning design specifications for shared instructional knowledge Canadian Journal of Learningand Technology 30 3
Conole G Dyke M Oliver M amp Seale J (2004) Mapping pedagogy and tools for effectivelearning design Computers amp Education 43 1ndash2 17ndash33
Dodero J M Aedo I amp Diacuteaz P (2002) Participative knowledge production of learning objectsfor e-books The Electronic Library 20 4 296ndash305
Friesen N (2005) Interoperability and learning objects an overview of e-learning standardiza-tion Interdisciplinary Journal of Knowledge and Learning Objects 1 1 23ndash31
Gruber T (1995) Towards principles for the design of ontologies used for knowledge sharingInternational Journal of HumanndashComputer Studies 43 56 907ndash928
Hatchuel A (2002) Towards design theory and expandable rationality the unfinished programof Herbert Simon Journal of Management and Governance 5 3ndash4 260ndash271
Holmes J (in press) Designing agents to support learning by explaining Computers and EducationIEEE Learning Technology Standards Committee (2002) Learning object metadata (LOM) Final
Draft Standard IEEE 1484121ndash2002IMS (2003) IMS Learning Design Information Model (Version 10) Final Specification 20
January 2003Kaptelinin V Kuutti K amp Bannon L (1995) Activity theory basic concepts and applications
In Blumenthalet al (Eds) Humanndashcomputer interaction Lecture Notes in Computer Science 1015189ndash201
Koper R (2004) Use of the semantic web to solve some basic problems in education increaseflexible distributed lifelong learning decrease teacherrsquos workload Journal of Interactive Mediain Education 2004 6
Lenat D (1995) CYC a large-scale investment in knowledge infrastructure Communications ofthe ACM 38 11 33ndash38
350 British Journal of Educational Technology Vol 37 No 3 2006
copy 2006 The Author Journal compilation copy 2006 British Educational Communications and Technology Agency
McGreal R (2004) Learning objects a practical definition International Journal of InstructionalTechnology and Distance Learning 1 9 21ndash32
Packer M amp Goicoechea J (2000) Sociocultural and constructivist theories of learningontology not just epistemology Educational Psychologist 35 4 227ndash241
Palmer J (2001) Fifty major thinkers on education London and New York RoutledgePolsani P R (2003) Use and abuse of reusable learning objects Journal of Digital Information 3
4 Rothwell W amp Kazanas H (1992) Mastering the instructional design process San Francisco CA
Jossey-BassSaacutenchez-Alonso S amp Sicilia M A (2004) Relationships and commitments in learning object
metadata In Proceedings of the 5th International Conference on Information Technology BasedHigher Education and Training ITHET 2004 Istanbul Turkey
Sicilia M A (2005) Ontology-Based Competency Management Infrastructures for theKnowledge-intensive Learning Organization In Lytras amp Naeve (Eds) Intelligent Learning Infra-structures in Knowledge Intensive Organizations A Semantic Web perspective IDEA USA pp 302ndash324
Sicilia M A amp Garciacutea E (2004) Fuzzy group models for adaptation in cooperative informationretrieval contexts Lecture Notes in Computer Science 2932 324ndash334
Sicilia M A amp Garciacutea E (2005) On the convergence of formal ontologies and standardizede-learning Journal of Distance Education Technologies 3 2 13ndash29
Sicilia M A amp Lytras M (2005) On the representation of change according to different ontol-ogies of learning International Journal of Learning and Change 1 1 66ndash79
Sicilia M A Lytras M Rodriacuteguez E amp Garciacutea E (in press) Integrating descriptions of knowl-edge management learning activities into large ontological structures a case study Data andKnowledge Engineering
Tennant M (1991) Expertise as a dimension of adult development New Education 13 1 49ndash55Vanderdonckt J (1999) Development milestones towards a tool for working with guidelines
Interacting with Computers 12 2 81ndash118Wand Y amp Weber R (1995) On the deep structure of information systems Information Systems
Journal 5 203ndash223Wiley D A (2001) The instructional use of learning objects Bloomington IN Association for
Educational Communications and Technology
Semantic learning designs 341
copy 2006 The Author Journal compilation copy 2006 British Educational Communications and Technology Agency
Several designers may participate in the same designing activity and formal frameworksfor collaborative learning resource creation (Dodero Aedo amp Diacuteaz 2002) can be inte-grated by clearly specifying which class of things can be the outcome of a designingprocess For the sake of flexibility the concept of LearningDesignPart is introduced as anall-encompassing category that includes each of the elements described in the previoussection This includes not only activities and roles but also objectives conditions andthe like Furthermore the granularity of the descriptions allows for different levels fromconsidering a unique designing activity for the whole package to recording each ele-ment of the learning design comprehensively Because oc_Actions are subsumed bythe concept of oc_Event considered as a lsquodynamic situation in which the state of theworld changesrsquo each edition of a metadata record or any other form of description maybe subject to be described The concept is flexible enough to be able to record fine-grained detailed decisions This of course may include different versions of the sameartefacts but this is an orthogonal aspect of configuration that is not relevant to ourpresent discussion
Figure 2 depicts the main elements of the loci to which rationales assumptions andguidelines will be attached
A LearningDesigner can be defined as the class of humans that performactivities of learning resource design The consideration of learning design as aunique human activity could be considered as restrictive of intelligent softwarefor the task Simply changing the oc_Designer class to a subconcept ofoc_IntelligentAgent could remove this but this entails a change in Open-Cyc knowledge base The concept oc_Designing is a subclass of the conceptoc_ThinkingAboutTheStateOfTheWorld which represents lsquothinking pro-cessesrsquo which include mental events such as planning evaluating or inventingThus the concept of design described so far should be understood in terms of the
Figure 2 Main elements of the ontological schema for the loci of design rationales
oc_performedBy modifies
342 British Journal of Educational Technology Vol 37 No 3 2006
copy 2006 The Author Journal compilation copy 2006 British Educational Communications and Technology Agency
concept of lsquoexpanded rationalityrsquo as described by Hatchuel (2002) integrating cre-ativity and unexpected expansions of the original requirements This precludes onto-logical definitions in which the problem space is completely bounded a priori Inconsequence a degree of openness is necessary to integrate different kinds of detailin description from fully described ones to others with shallower semantics eg someproviding only references to generic assumptions
Theoretical frameworksWe will make the distinction that guidelines (as further defined) are the propositions thatdirectly predicate about the validity or appropriateness of learning design parts whileother theoretical propositions about learning that are not directly saying anythingabout the design of digital learning resources are considered as assumptions followingthe general sense provided in the Merriam-Webster dictionary lsquoa fact or statement (asa proposition axiom postulate or notion) taken for grantedrsquo
The fact that some assumption hypotheses or proposed law about learning is sustainedby a broader empirical base or is better sustained by rational means is out of the scopeof this work because here we focus on design practice (even though it is of an obviousinterest for researchers) In principle even unorthodox positions may be sustainedbecause the mechanism of microtheories and their inclusion mechanism allows forgreat flexibility in modelling them
Table 2 provides some examples of theoretical assumptions and related guidelines forlsquoextremersquo archetypical standpoints on learning These are not intended to reflect actualcommon positions but they are provided as illustrative examples of radical positionsthat could be modelled with our ontological schema
The examples provided in Table 2 go a step further than toolkits in Conole et al (2004)in making the design rationale explicit Nonetheless in many cases lsquocommon sensersquo isapplied to design and a shallower approach to specification could be used instead egproviding simply links to the theoretical standpoints used It should be noted also thatthe resulting guidelines vary greatly in their level of concreteness and some of themmay be considered as mandatory from some standpoints while others are mere generalrepresentations This would lead to different computational strategies
Guidelines and rulesThe WordNet (httpwordnetprincetonedu) thesaurus defines guidelines as lsquoguidancerelative to setting standards or determining a course of actionrsquo or lsquoa rule or principlethat provides guidance to appropriate behaviorrsquo In our case guidelines are statementsthat serve as criteria for decision in designing learning Given that e-learning designis still far from being a mature discipline guidelinesmdashas instances of recordedknowledgemdashcan be characterised as oc_NormativeSpecifications becausethey describe how learning resource design activities and their outcomes lsquoshould bersquoin a propositional form Propositions lsquohave some truth value in some context or
Semantic learning designs 343
copy 2006 The Author Journal compilation copy 2006 British Educational Communications and Technology Agency
ldquoworldrdquo rsquo Furthermore propositions lsquoare assumed to be representable (at least in prin-ciple) by a sentence in some formal or natural languagersquo This raises the need for alanguage that expresses guidelines in a general sense There are two main options forsuch representation be them formal logical sentences or reified guidelines
Formal logical sentence representation entails the formalisation through sentences ofthe constraints of the designs For example if lsquoevery learning resource must be activersquoaccording to some theoretical standpoint (eg one that fosters reflective learning) thissentence may be represented as an axiom and put into the appropriate separate micro-theory as the following CycL fragment
($forAll X ($implies
($isa X $LearningObject)
($resourceType X $Active))
Then resources that are not active are simply excluded from the notion itself This is astrong position in the sense that precludes some views on the learning design process
Table 2 Some examples of theoretical assumptions and guidelines
Archetypicalstandpoint Assumptions Guidelines
Instructivist(Behaviouristic)
bull Learning occurs through association and reinforcement
bull The results of learning can be appreciated through observable outcomes
(GB1) Every instance of oc-Learning entails theexposure to some LearningObjectthat produces the association (GB2) Learning Activitysequences should provide assessmentactivities that confirm theassociations that should have beenestablished so far
Constructivist bull Learning proceeds by building mental structures
(GC1) Agentrsquos mental structuresshould be represented(GC2) These structures should bepre-assessed in order to becomparable with the resulting ones(GC3) Design Activities arerecommended(GC4) Contrast Activities arerecommended
Sociocultural bull Learning occurs through social participatio
(GSC1) Every instance ofoc-Learning entails theinteraction of severaloc-IntelligentAgents asoc-Actors
344 British Journal of Educational Technology Vol 37 No 3 2006
copy 2006 The Author Journal compilation copy 2006 British Educational Communications and Technology Agency
Another typical example may be that of a sociocultural guideline that preclude Methodsthat do not entail multirole activities that provide access to social interaction ServicesThis approach to guidelines provides the best framework for automation and wouldenable the creation of microtheories that represent clearly formulated views onlearning Nonetheless there are two major problems for the use of this option as theunique mechanism for encoding design rationales
1 There is no clear-cut distinction between models and theories of learning becausethey come from a history of schools of thought with interrelations (Palmer 2001)This would lead to the need for the representation of dozens of variants reflectingpositions that only diverge in some aspects
2 Many guidelines cannot be represented as logical sentences because theyrepresent lsquorecommendationsrsquo or vague requirements eg see (GC3) or (GC4)(Table 2)
These issues suggest that a combination of formal sentences with reified guidelinesprovide the best option for flexibility The representation of guidelines has beenstudied in the field of HumanndashComputer-Interaction (HCI) in the last decade(Vanderdonckt 1999) and the main philosophy of such representations remainsvalid in the learning design domain The major challenge for such representations isthat of providing computational semantics that could be used for the implementationof software support but this is difficult to attain in all the cases Thus the concept oflearning design guideline will provide two kinds of representations general andschema-specific
The general use of instances of oc_NormativeSpecification allows the model-ling of lsquoconceptual works that describe how something should bersquo This may includerepresentation of known normative representations or procedures For example onemay refer to a concrete learning design model as that derived from Bannister (1981)which states that creativity cycles follow an initial phase of brainstorming (idea cap-ture) followed by a phase in which key issues are focused into an operation strategy(idea development) and a third phase in which project control takes place through alsquorecursive cyclersquo (operational management) The reference could be made by pointing tothe academic reference describing it as a conceptual work or it could be described asan Activity-Spec modelled explicitly inside the ontology As another examplesome detailed guidelines could be modelled by the normative specification sub-category of oc-Instructions defined as lsquoa sequence of tasks to be performed byan agentrsquo
Schema-specific guidelines restrict the freedom to represent criteria as conceptual workinstances to a concrete and commonly agreed ontology that determine what can be saidabout how learning designs should be according to a theoretical standpoint or frame-work These build on an activityndashrole ontology as the one described in the previoussection and add a prescription on the kind of lsquoshould-bersquo predications that are allow-able Table 3 describes examples of a minimal language that could be used for thatpurpose
Semantic learning designs 345
copy 2006 The Author Journal compilation copy 2006 British Educational Communications and Technology Agency
In the examples CAT represents a category of individual expressed through a conceptor a logical formula and the sentence is a synthetic way of expressing some recom-mended or mandatory constraint over such category Other languages could be engi-neered in further research but the one described can be directly expressed in OpenCycand combined with software modules that check or enforce them in some way Notethat this representation does not use direct OpenCyc sentences so that reification is usedinstead of direct logical sentences It is not the aim of this paper to determine the mostpowerful and flexible representation language but only set a foundation for furtherspecification efforts
Analysis and comparison of two theoretical semantic frameworksAs an assessment of the representational capabilities of the scheme just described acase study of representation is sketched in this section We will use the learning designtoolkit described by Conole et al (2004) as the underlying terminology even thoughother frameworks could be used instead
Two rather diverging positions for design will be used as an illustration Table 4 sum-marises the two design scenarios
Table 3 Possible language for use in learning designs
Sentence type Potential uses Example
RecommendationshouldBe (CATSENT)
Recommend kinds ofactivities
Design activities are recommended (GC3)shouldBe(Method subEvent (CATDesignActivity))
ObligationmustBe (CATSENT)
Some activity typesmust follow others
Assessments must follow each expositionto contents (GD1) mustBe(and(ExpositiveActivityIndividualActivity)startAfterEndingOf(CAT
AssessmentActivity)
Activities must havesome characteristic
Every activity must entail interactionmustBe(LDActivity(usedIn(CAT
InteractiveService)))Every activity must provide acontent-oriented learning object forindividual use (similar to GB1)mustBe(LDActivity(and (usedIn(CAT
LearningObject)(learnerRoleNumber(1))))
Some structure shouldbe enforced in methods
Methods should have a sequential actstructuremustBe(Methodnot ((subEvent(CAT Play))))
CAT = category GC3 = Constructivist Guideline 3 (see Table 2) GD1 = Guideline 1
346 British Journal of Educational Technology Vol 37 No 3 2006
copy 2006 The Author Journal compilation copy 2006 British Educational Communications and Technology Agency
The following subsections sketch the main outcomes of an example learning designprocess for both scenarios assuming that the process described by Conole et al (2004)is used in both situations
Differences in modelling the design processThe forces driving the design process in both scenarios are fairly diverging The designof Scenario A will typically be constrained by a clear target specified in terms of anaggregated competency level while in Scenario B there are no concrete competenciesto be acquired and the main objective of the design activity would be twofold On theone hand the activities are targeted at promoting interchange of ideas in an attemptto stimulate casual convergence of interest which is the main value of the culture ofthe special interest group On the other hand the activities have the generic objectiveof strengthening social ties which is an important element of cohesion for the under-taking of joint projects
Table 5 describes the main inputs and possible activities and their objectives at a highlevel
The changes intended to be achieved in each of the scenarios are stated differently witha formal definition in (A) and a shallow cognitive and social representation in the case
Table 4 Learning design scenarios
Scenario DescriptionClassification according to
Conole et al (2004) toolkit
(A) Systemic-Behaviourist A software development company attempts to designlearning units about newtechnologies or softwareproducts for its developerstaff The design is drivenby a cycle of lsquocompetencygap analysisrsquo that modelsorganisational learning in terms of the aggregated competencies of the employeesmdashsimilar to theone described by Sicilia(2005)
Focuses on the individual where a body of informationforms the basis and rawmaterial for learning Learning is mainly considered as nonreflective ie based on skill learning and memorisation
(B) Sociocultural A special interest group ona concrete research topicaims at developing the knowledge of its membersby sharing experiencesand discussing on themesof common interest
Focuses on the social whereexperience is the main source of learning Conscious reflection on experience is transformed into learning
Semantic learning designs 347
copy 2006 The Author Journal compilation copy 2006 British Educational Communications and Technology Agency
of (B) In addition case (B) requires a model of social network that is largely irrelevantin case (A)
Differences in recording activity rationalesThe approach for the recording of design rationales also differs because of the differentcontexts Table 6 summarises the main elements for each of the scenarios
Conclusions and directions for researchCurrent technology and practice related to the description of reusable digital learningresources has resulted in techniques that allow the description of learning activities of
Table 5 Inputs activities and objectives of two learning design scenarios
Scenario Design inputs Activities and their objectives
(A) Systemic-Behaviourist A formal competency recordfor employees and competencyrequirements expressed in terms of amounts of required competencies (assuming a shared measurement scale) Such requirements are actually derived from strategy or from the contingent requirements of upcoming projects
Competency instances are used to measure the levelof Competency-Spec for the required competencies (or their constituents) A formal match considering the agenda of employees is carried out for determining the target learners LearningObjects forindividual use are selected asthe main elements of theactivities and the criteria forselection are again a match ofthe objectives stated inmetadata (as stated incategory 9 of Learning Object MetadatamdashClassification) and thecompetencies required
(B) Sociocultural The expertise of the special interest group members as stated in the concepts addressed by their publicationsprojects or other activities
Informal group activities are the main approach applicablesince there is not a predefinedgroup structure but a cultureof sharing and socialisation The results of the sharing of ideas can be expressed in terms of oc-knowsAboutpredicates and social ties will be modelled through positive valence FOAF-knows(httpxmlnscomfoaf01)predicates with an added numerical value representingstrength
348 British Journal of Educational Technology Vol 37 No 3 2006
copy 2006 The Author Journal compilation copy 2006 British Educational Communications and Technology Agency
a complex kind including their intended participants educational objectives and exe-cution constraints The representation and recording of the assumptions and guidelinesthat were considered in the creation of such resources would extend current techniquesto capture the rationale of the design itself This in turn will eventually result in theavailability of a large base of empirical data available in which theoretical assumptionsor rules are explicitly connected to the resulting designs as these could be linked withthe actual results and trace of the learning processes that used them The resulting bodyof evidence would represent a critical asset in the development of the theory of learningmediated through computers
This paper has described the first attempt to extend existing learning technology modelswith a schema explicitly targeted at describing learning design rationales The OpenCycknowledge base has been used as the semantic reference from which the ontologicalschema has been specified Concretely an activity-based schema based on IMS LD hasbeen described and theoretical assumptions and codified guidelines have been intro-duced The flexibility of the representation schema has been analysed through therepresentation of two stereotyped theoretical positions
Two main research directions should continue the work described here On the onehand further ontological engineering effort is required to come up with a more com-prehensive and rich schema for learning design rationales On the other hand tool
Table 6 Main elements of the two learning design scenarios
Scenario Approach Examples
(A) Systemic-Behaviourist Formal logical sentencesare used to guarantee that every design followsan organisationalobjective
It is required that the objective of every Method is attached to a Competency-Spec instance as objective
Formal assessment is a requirement foraccountability purposes
Guideline (GD1) is attached to everyDesigningLearningResource instance that produces an Act
(B) Sociocultural The interaction in informal settings can beused to derive social models (Sicilia amp Garciacutea2004)
The interactions in discussion canbe used to measure the degree of lsquoproximityrsquo between members Therationale of the activities can bejustified in terms of SocialTies inside SocialNetworks Further the focus of each groupactivity should be clearly stated anda combination of the increase in collective knowledge and the strengthening of social relations can be used for assessment
Semantic learning designs 349
copy 2006 The Author Journal compilation copy 2006 British Educational Communications and Technology Agency
implementation and record gathering efforts would be required to attain the ultimateobjective of the ontological definitions described so far With such effort an unprece-dented research material could be available in some years providing improved oppor-tunities for scientific inquiry on design as an activity regarding pedagogy
Other potential directions for research may be the description of interpretations of thedesign intentions (for which no rationale was explicitly recorded) the mining or detec-tion of patterns as related to rationale description or the detailed use of theoreticalrationales in agents that support the learning process which represent a category ofrelevant learning support software (Holmes in press)
ReferencesAllert H (2004) Coherent social systems for learningmdashan approach for contextualized and
community-centred metadata Journal of Interactive Media in Education 2004 1Baader F Calvanese D McGuinness D Nardi D amp Patel-Schneider P (Eds) (2003) The
description logic handbook Theory implementation and applications Cambridge CambridgeUniversity Press
Bannister D (1981) Personal construct theory and research method In P Reason amp J Rowan(Eds) Human inquiry a sourcebook of new paradigm research (pp 191ndash200) Chichester UKJohn Wiley and Sons
Berners-Lee T Hendler J amp Lassila O (2001) The semantic web Scientific American 284 534ndash43
Brown A amp Voltz B (2005) Elements of effective e-learning design International Review ofResearch in Open and Distance Learning 6 1
Buzza D Bean D Harrigan K amp Carey T (2005) Learning design repositories adaptinglearning design specifications for shared instructional knowledge Canadian Journal of Learningand Technology 30 3
Conole G Dyke M Oliver M amp Seale J (2004) Mapping pedagogy and tools for effectivelearning design Computers amp Education 43 1ndash2 17ndash33
Dodero J M Aedo I amp Diacuteaz P (2002) Participative knowledge production of learning objectsfor e-books The Electronic Library 20 4 296ndash305
Friesen N (2005) Interoperability and learning objects an overview of e-learning standardiza-tion Interdisciplinary Journal of Knowledge and Learning Objects 1 1 23ndash31
Gruber T (1995) Towards principles for the design of ontologies used for knowledge sharingInternational Journal of HumanndashComputer Studies 43 56 907ndash928
Hatchuel A (2002) Towards design theory and expandable rationality the unfinished programof Herbert Simon Journal of Management and Governance 5 3ndash4 260ndash271
Holmes J (in press) Designing agents to support learning by explaining Computers and EducationIEEE Learning Technology Standards Committee (2002) Learning object metadata (LOM) Final
Draft Standard IEEE 1484121ndash2002IMS (2003) IMS Learning Design Information Model (Version 10) Final Specification 20
January 2003Kaptelinin V Kuutti K amp Bannon L (1995) Activity theory basic concepts and applications
In Blumenthalet al (Eds) Humanndashcomputer interaction Lecture Notes in Computer Science 1015189ndash201
Koper R (2004) Use of the semantic web to solve some basic problems in education increaseflexible distributed lifelong learning decrease teacherrsquos workload Journal of Interactive Mediain Education 2004 6
Lenat D (1995) CYC a large-scale investment in knowledge infrastructure Communications ofthe ACM 38 11 33ndash38
350 British Journal of Educational Technology Vol 37 No 3 2006
copy 2006 The Author Journal compilation copy 2006 British Educational Communications and Technology Agency
McGreal R (2004) Learning objects a practical definition International Journal of InstructionalTechnology and Distance Learning 1 9 21ndash32
Packer M amp Goicoechea J (2000) Sociocultural and constructivist theories of learningontology not just epistemology Educational Psychologist 35 4 227ndash241
Palmer J (2001) Fifty major thinkers on education London and New York RoutledgePolsani P R (2003) Use and abuse of reusable learning objects Journal of Digital Information 3
4 Rothwell W amp Kazanas H (1992) Mastering the instructional design process San Francisco CA
Jossey-BassSaacutenchez-Alonso S amp Sicilia M A (2004) Relationships and commitments in learning object
metadata In Proceedings of the 5th International Conference on Information Technology BasedHigher Education and Training ITHET 2004 Istanbul Turkey
Sicilia M A (2005) Ontology-Based Competency Management Infrastructures for theKnowledge-intensive Learning Organization In Lytras amp Naeve (Eds) Intelligent Learning Infra-structures in Knowledge Intensive Organizations A Semantic Web perspective IDEA USA pp 302ndash324
Sicilia M A amp Garciacutea E (2004) Fuzzy group models for adaptation in cooperative informationretrieval contexts Lecture Notes in Computer Science 2932 324ndash334
Sicilia M A amp Garciacutea E (2005) On the convergence of formal ontologies and standardizede-learning Journal of Distance Education Technologies 3 2 13ndash29
Sicilia M A amp Lytras M (2005) On the representation of change according to different ontol-ogies of learning International Journal of Learning and Change 1 1 66ndash79
Sicilia M A Lytras M Rodriacuteguez E amp Garciacutea E (in press) Integrating descriptions of knowl-edge management learning activities into large ontological structures a case study Data andKnowledge Engineering
Tennant M (1991) Expertise as a dimension of adult development New Education 13 1 49ndash55Vanderdonckt J (1999) Development milestones towards a tool for working with guidelines
Interacting with Computers 12 2 81ndash118Wand Y amp Weber R (1995) On the deep structure of information systems Information Systems
Journal 5 203ndash223Wiley D A (2001) The instructional use of learning objects Bloomington IN Association for
Educational Communications and Technology
342 British Journal of Educational Technology Vol 37 No 3 2006
copy 2006 The Author Journal compilation copy 2006 British Educational Communications and Technology Agency
concept of lsquoexpanded rationalityrsquo as described by Hatchuel (2002) integrating cre-ativity and unexpected expansions of the original requirements This precludes onto-logical definitions in which the problem space is completely bounded a priori Inconsequence a degree of openness is necessary to integrate different kinds of detailin description from fully described ones to others with shallower semantics eg someproviding only references to generic assumptions
Theoretical frameworksWe will make the distinction that guidelines (as further defined) are the propositions thatdirectly predicate about the validity or appropriateness of learning design parts whileother theoretical propositions about learning that are not directly saying anythingabout the design of digital learning resources are considered as assumptions followingthe general sense provided in the Merriam-Webster dictionary lsquoa fact or statement (asa proposition axiom postulate or notion) taken for grantedrsquo
The fact that some assumption hypotheses or proposed law about learning is sustainedby a broader empirical base or is better sustained by rational means is out of the scopeof this work because here we focus on design practice (even though it is of an obviousinterest for researchers) In principle even unorthodox positions may be sustainedbecause the mechanism of microtheories and their inclusion mechanism allows forgreat flexibility in modelling them
Table 2 provides some examples of theoretical assumptions and related guidelines forlsquoextremersquo archetypical standpoints on learning These are not intended to reflect actualcommon positions but they are provided as illustrative examples of radical positionsthat could be modelled with our ontological schema
The examples provided in Table 2 go a step further than toolkits in Conole et al (2004)in making the design rationale explicit Nonetheless in many cases lsquocommon sensersquo isapplied to design and a shallower approach to specification could be used instead egproviding simply links to the theoretical standpoints used It should be noted also thatthe resulting guidelines vary greatly in their level of concreteness and some of themmay be considered as mandatory from some standpoints while others are mere generalrepresentations This would lead to different computational strategies
Guidelines and rulesThe WordNet (httpwordnetprincetonedu) thesaurus defines guidelines as lsquoguidancerelative to setting standards or determining a course of actionrsquo or lsquoa rule or principlethat provides guidance to appropriate behaviorrsquo In our case guidelines are statementsthat serve as criteria for decision in designing learning Given that e-learning designis still far from being a mature discipline guidelinesmdashas instances of recordedknowledgemdashcan be characterised as oc_NormativeSpecifications becausethey describe how learning resource design activities and their outcomes lsquoshould bersquoin a propositional form Propositions lsquohave some truth value in some context or
Semantic learning designs 343
copy 2006 The Author Journal compilation copy 2006 British Educational Communications and Technology Agency
ldquoworldrdquo rsquo Furthermore propositions lsquoare assumed to be representable (at least in prin-ciple) by a sentence in some formal or natural languagersquo This raises the need for alanguage that expresses guidelines in a general sense There are two main options forsuch representation be them formal logical sentences or reified guidelines
Formal logical sentence representation entails the formalisation through sentences ofthe constraints of the designs For example if lsquoevery learning resource must be activersquoaccording to some theoretical standpoint (eg one that fosters reflective learning) thissentence may be represented as an axiom and put into the appropriate separate micro-theory as the following CycL fragment
($forAll X ($implies
($isa X $LearningObject)
($resourceType X $Active))
Then resources that are not active are simply excluded from the notion itself This is astrong position in the sense that precludes some views on the learning design process
Table 2 Some examples of theoretical assumptions and guidelines
Archetypicalstandpoint Assumptions Guidelines
Instructivist(Behaviouristic)
bull Learning occurs through association and reinforcement
bull The results of learning can be appreciated through observable outcomes
(GB1) Every instance of oc-Learning entails theexposure to some LearningObjectthat produces the association (GB2) Learning Activitysequences should provide assessmentactivities that confirm theassociations that should have beenestablished so far
Constructivist bull Learning proceeds by building mental structures
(GC1) Agentrsquos mental structuresshould be represented(GC2) These structures should bepre-assessed in order to becomparable with the resulting ones(GC3) Design Activities arerecommended(GC4) Contrast Activities arerecommended
Sociocultural bull Learning occurs through social participatio
(GSC1) Every instance ofoc-Learning entails theinteraction of severaloc-IntelligentAgents asoc-Actors
344 British Journal of Educational Technology Vol 37 No 3 2006
copy 2006 The Author Journal compilation copy 2006 British Educational Communications and Technology Agency
Another typical example may be that of a sociocultural guideline that preclude Methodsthat do not entail multirole activities that provide access to social interaction ServicesThis approach to guidelines provides the best framework for automation and wouldenable the creation of microtheories that represent clearly formulated views onlearning Nonetheless there are two major problems for the use of this option as theunique mechanism for encoding design rationales
1 There is no clear-cut distinction between models and theories of learning becausethey come from a history of schools of thought with interrelations (Palmer 2001)This would lead to the need for the representation of dozens of variants reflectingpositions that only diverge in some aspects
2 Many guidelines cannot be represented as logical sentences because theyrepresent lsquorecommendationsrsquo or vague requirements eg see (GC3) or (GC4)(Table 2)
These issues suggest that a combination of formal sentences with reified guidelinesprovide the best option for flexibility The representation of guidelines has beenstudied in the field of HumanndashComputer-Interaction (HCI) in the last decade(Vanderdonckt 1999) and the main philosophy of such representations remainsvalid in the learning design domain The major challenge for such representations isthat of providing computational semantics that could be used for the implementationof software support but this is difficult to attain in all the cases Thus the concept oflearning design guideline will provide two kinds of representations general andschema-specific
The general use of instances of oc_NormativeSpecification allows the model-ling of lsquoconceptual works that describe how something should bersquo This may includerepresentation of known normative representations or procedures For example onemay refer to a concrete learning design model as that derived from Bannister (1981)which states that creativity cycles follow an initial phase of brainstorming (idea cap-ture) followed by a phase in which key issues are focused into an operation strategy(idea development) and a third phase in which project control takes place through alsquorecursive cyclersquo (operational management) The reference could be made by pointing tothe academic reference describing it as a conceptual work or it could be described asan Activity-Spec modelled explicitly inside the ontology As another examplesome detailed guidelines could be modelled by the normative specification sub-category of oc-Instructions defined as lsquoa sequence of tasks to be performed byan agentrsquo
Schema-specific guidelines restrict the freedom to represent criteria as conceptual workinstances to a concrete and commonly agreed ontology that determine what can be saidabout how learning designs should be according to a theoretical standpoint or frame-work These build on an activityndashrole ontology as the one described in the previoussection and add a prescription on the kind of lsquoshould-bersquo predications that are allow-able Table 3 describes examples of a minimal language that could be used for thatpurpose
Semantic learning designs 345
copy 2006 The Author Journal compilation copy 2006 British Educational Communications and Technology Agency
In the examples CAT represents a category of individual expressed through a conceptor a logical formula and the sentence is a synthetic way of expressing some recom-mended or mandatory constraint over such category Other languages could be engi-neered in further research but the one described can be directly expressed in OpenCycand combined with software modules that check or enforce them in some way Notethat this representation does not use direct OpenCyc sentences so that reification is usedinstead of direct logical sentences It is not the aim of this paper to determine the mostpowerful and flexible representation language but only set a foundation for furtherspecification efforts
Analysis and comparison of two theoretical semantic frameworksAs an assessment of the representational capabilities of the scheme just described acase study of representation is sketched in this section We will use the learning designtoolkit described by Conole et al (2004) as the underlying terminology even thoughother frameworks could be used instead
Two rather diverging positions for design will be used as an illustration Table 4 sum-marises the two design scenarios
Table 3 Possible language for use in learning designs
Sentence type Potential uses Example
RecommendationshouldBe (CATSENT)
Recommend kinds ofactivities
Design activities are recommended (GC3)shouldBe(Method subEvent (CATDesignActivity))
ObligationmustBe (CATSENT)
Some activity typesmust follow others
Assessments must follow each expositionto contents (GD1) mustBe(and(ExpositiveActivityIndividualActivity)startAfterEndingOf(CAT
AssessmentActivity)
Activities must havesome characteristic
Every activity must entail interactionmustBe(LDActivity(usedIn(CAT
InteractiveService)))Every activity must provide acontent-oriented learning object forindividual use (similar to GB1)mustBe(LDActivity(and (usedIn(CAT
LearningObject)(learnerRoleNumber(1))))
Some structure shouldbe enforced in methods
Methods should have a sequential actstructuremustBe(Methodnot ((subEvent(CAT Play))))
CAT = category GC3 = Constructivist Guideline 3 (see Table 2) GD1 = Guideline 1
346 British Journal of Educational Technology Vol 37 No 3 2006
copy 2006 The Author Journal compilation copy 2006 British Educational Communications and Technology Agency
The following subsections sketch the main outcomes of an example learning designprocess for both scenarios assuming that the process described by Conole et al (2004)is used in both situations
Differences in modelling the design processThe forces driving the design process in both scenarios are fairly diverging The designof Scenario A will typically be constrained by a clear target specified in terms of anaggregated competency level while in Scenario B there are no concrete competenciesto be acquired and the main objective of the design activity would be twofold On theone hand the activities are targeted at promoting interchange of ideas in an attemptto stimulate casual convergence of interest which is the main value of the culture ofthe special interest group On the other hand the activities have the generic objectiveof strengthening social ties which is an important element of cohesion for the under-taking of joint projects
Table 5 describes the main inputs and possible activities and their objectives at a highlevel
The changes intended to be achieved in each of the scenarios are stated differently witha formal definition in (A) and a shallow cognitive and social representation in the case
Table 4 Learning design scenarios
Scenario DescriptionClassification according to
Conole et al (2004) toolkit
(A) Systemic-Behaviourist A software development company attempts to designlearning units about newtechnologies or softwareproducts for its developerstaff The design is drivenby a cycle of lsquocompetencygap analysisrsquo that modelsorganisational learning in terms of the aggregated competencies of the employeesmdashsimilar to theone described by Sicilia(2005)
Focuses on the individual where a body of informationforms the basis and rawmaterial for learning Learning is mainly considered as nonreflective ie based on skill learning and memorisation
(B) Sociocultural A special interest group ona concrete research topicaims at developing the knowledge of its membersby sharing experiencesand discussing on themesof common interest
Focuses on the social whereexperience is the main source of learning Conscious reflection on experience is transformed into learning
Semantic learning designs 347
copy 2006 The Author Journal compilation copy 2006 British Educational Communications and Technology Agency
of (B) In addition case (B) requires a model of social network that is largely irrelevantin case (A)
Differences in recording activity rationalesThe approach for the recording of design rationales also differs because of the differentcontexts Table 6 summarises the main elements for each of the scenarios
Conclusions and directions for researchCurrent technology and practice related to the description of reusable digital learningresources has resulted in techniques that allow the description of learning activities of
Table 5 Inputs activities and objectives of two learning design scenarios
Scenario Design inputs Activities and their objectives
(A) Systemic-Behaviourist A formal competency recordfor employees and competencyrequirements expressed in terms of amounts of required competencies (assuming a shared measurement scale) Such requirements are actually derived from strategy or from the contingent requirements of upcoming projects
Competency instances are used to measure the levelof Competency-Spec for the required competencies (or their constituents) A formal match considering the agenda of employees is carried out for determining the target learners LearningObjects forindividual use are selected asthe main elements of theactivities and the criteria forselection are again a match ofthe objectives stated inmetadata (as stated incategory 9 of Learning Object MetadatamdashClassification) and thecompetencies required
(B) Sociocultural The expertise of the special interest group members as stated in the concepts addressed by their publicationsprojects or other activities
Informal group activities are the main approach applicablesince there is not a predefinedgroup structure but a cultureof sharing and socialisation The results of the sharing of ideas can be expressed in terms of oc-knowsAboutpredicates and social ties will be modelled through positive valence FOAF-knows(httpxmlnscomfoaf01)predicates with an added numerical value representingstrength
348 British Journal of Educational Technology Vol 37 No 3 2006
copy 2006 The Author Journal compilation copy 2006 British Educational Communications and Technology Agency
a complex kind including their intended participants educational objectives and exe-cution constraints The representation and recording of the assumptions and guidelinesthat were considered in the creation of such resources would extend current techniquesto capture the rationale of the design itself This in turn will eventually result in theavailability of a large base of empirical data available in which theoretical assumptionsor rules are explicitly connected to the resulting designs as these could be linked withthe actual results and trace of the learning processes that used them The resulting bodyof evidence would represent a critical asset in the development of the theory of learningmediated through computers
This paper has described the first attempt to extend existing learning technology modelswith a schema explicitly targeted at describing learning design rationales The OpenCycknowledge base has been used as the semantic reference from which the ontologicalschema has been specified Concretely an activity-based schema based on IMS LD hasbeen described and theoretical assumptions and codified guidelines have been intro-duced The flexibility of the representation schema has been analysed through therepresentation of two stereotyped theoretical positions
Two main research directions should continue the work described here On the onehand further ontological engineering effort is required to come up with a more com-prehensive and rich schema for learning design rationales On the other hand tool
Table 6 Main elements of the two learning design scenarios
Scenario Approach Examples
(A) Systemic-Behaviourist Formal logical sentencesare used to guarantee that every design followsan organisationalobjective
It is required that the objective of every Method is attached to a Competency-Spec instance as objective
Formal assessment is a requirement foraccountability purposes
Guideline (GD1) is attached to everyDesigningLearningResource instance that produces an Act
(B) Sociocultural The interaction in informal settings can beused to derive social models (Sicilia amp Garciacutea2004)
The interactions in discussion canbe used to measure the degree of lsquoproximityrsquo between members Therationale of the activities can bejustified in terms of SocialTies inside SocialNetworks Further the focus of each groupactivity should be clearly stated anda combination of the increase in collective knowledge and the strengthening of social relations can be used for assessment
Semantic learning designs 349
copy 2006 The Author Journal compilation copy 2006 British Educational Communications and Technology Agency
implementation and record gathering efforts would be required to attain the ultimateobjective of the ontological definitions described so far With such effort an unprece-dented research material could be available in some years providing improved oppor-tunities for scientific inquiry on design as an activity regarding pedagogy
Other potential directions for research may be the description of interpretations of thedesign intentions (for which no rationale was explicitly recorded) the mining or detec-tion of patterns as related to rationale description or the detailed use of theoreticalrationales in agents that support the learning process which represent a category ofrelevant learning support software (Holmes in press)
ReferencesAllert H (2004) Coherent social systems for learningmdashan approach for contextualized and
community-centred metadata Journal of Interactive Media in Education 2004 1Baader F Calvanese D McGuinness D Nardi D amp Patel-Schneider P (Eds) (2003) The
description logic handbook Theory implementation and applications Cambridge CambridgeUniversity Press
Bannister D (1981) Personal construct theory and research method In P Reason amp J Rowan(Eds) Human inquiry a sourcebook of new paradigm research (pp 191ndash200) Chichester UKJohn Wiley and Sons
Berners-Lee T Hendler J amp Lassila O (2001) The semantic web Scientific American 284 534ndash43
Brown A amp Voltz B (2005) Elements of effective e-learning design International Review ofResearch in Open and Distance Learning 6 1
Buzza D Bean D Harrigan K amp Carey T (2005) Learning design repositories adaptinglearning design specifications for shared instructional knowledge Canadian Journal of Learningand Technology 30 3
Conole G Dyke M Oliver M amp Seale J (2004) Mapping pedagogy and tools for effectivelearning design Computers amp Education 43 1ndash2 17ndash33
Dodero J M Aedo I amp Diacuteaz P (2002) Participative knowledge production of learning objectsfor e-books The Electronic Library 20 4 296ndash305
Friesen N (2005) Interoperability and learning objects an overview of e-learning standardiza-tion Interdisciplinary Journal of Knowledge and Learning Objects 1 1 23ndash31
Gruber T (1995) Towards principles for the design of ontologies used for knowledge sharingInternational Journal of HumanndashComputer Studies 43 56 907ndash928
Hatchuel A (2002) Towards design theory and expandable rationality the unfinished programof Herbert Simon Journal of Management and Governance 5 3ndash4 260ndash271
Holmes J (in press) Designing agents to support learning by explaining Computers and EducationIEEE Learning Technology Standards Committee (2002) Learning object metadata (LOM) Final
Draft Standard IEEE 1484121ndash2002IMS (2003) IMS Learning Design Information Model (Version 10) Final Specification 20
January 2003Kaptelinin V Kuutti K amp Bannon L (1995) Activity theory basic concepts and applications
In Blumenthalet al (Eds) Humanndashcomputer interaction Lecture Notes in Computer Science 1015189ndash201
Koper R (2004) Use of the semantic web to solve some basic problems in education increaseflexible distributed lifelong learning decrease teacherrsquos workload Journal of Interactive Mediain Education 2004 6
Lenat D (1995) CYC a large-scale investment in knowledge infrastructure Communications ofthe ACM 38 11 33ndash38
350 British Journal of Educational Technology Vol 37 No 3 2006
copy 2006 The Author Journal compilation copy 2006 British Educational Communications and Technology Agency
McGreal R (2004) Learning objects a practical definition International Journal of InstructionalTechnology and Distance Learning 1 9 21ndash32
Packer M amp Goicoechea J (2000) Sociocultural and constructivist theories of learningontology not just epistemology Educational Psychologist 35 4 227ndash241
Palmer J (2001) Fifty major thinkers on education London and New York RoutledgePolsani P R (2003) Use and abuse of reusable learning objects Journal of Digital Information 3
4 Rothwell W amp Kazanas H (1992) Mastering the instructional design process San Francisco CA
Jossey-BassSaacutenchez-Alonso S amp Sicilia M A (2004) Relationships and commitments in learning object
metadata In Proceedings of the 5th International Conference on Information Technology BasedHigher Education and Training ITHET 2004 Istanbul Turkey
Sicilia M A (2005) Ontology-Based Competency Management Infrastructures for theKnowledge-intensive Learning Organization In Lytras amp Naeve (Eds) Intelligent Learning Infra-structures in Knowledge Intensive Organizations A Semantic Web perspective IDEA USA pp 302ndash324
Sicilia M A amp Garciacutea E (2004) Fuzzy group models for adaptation in cooperative informationretrieval contexts Lecture Notes in Computer Science 2932 324ndash334
Sicilia M A amp Garciacutea E (2005) On the convergence of formal ontologies and standardizede-learning Journal of Distance Education Technologies 3 2 13ndash29
Sicilia M A amp Lytras M (2005) On the representation of change according to different ontol-ogies of learning International Journal of Learning and Change 1 1 66ndash79
Sicilia M A Lytras M Rodriacuteguez E amp Garciacutea E (in press) Integrating descriptions of knowl-edge management learning activities into large ontological structures a case study Data andKnowledge Engineering
Tennant M (1991) Expertise as a dimension of adult development New Education 13 1 49ndash55Vanderdonckt J (1999) Development milestones towards a tool for working with guidelines
Interacting with Computers 12 2 81ndash118Wand Y amp Weber R (1995) On the deep structure of information systems Information Systems
Journal 5 203ndash223Wiley D A (2001) The instructional use of learning objects Bloomington IN Association for
Educational Communications and Technology
Semantic learning designs 343
copy 2006 The Author Journal compilation copy 2006 British Educational Communications and Technology Agency
ldquoworldrdquo rsquo Furthermore propositions lsquoare assumed to be representable (at least in prin-ciple) by a sentence in some formal or natural languagersquo This raises the need for alanguage that expresses guidelines in a general sense There are two main options forsuch representation be them formal logical sentences or reified guidelines
Formal logical sentence representation entails the formalisation through sentences ofthe constraints of the designs For example if lsquoevery learning resource must be activersquoaccording to some theoretical standpoint (eg one that fosters reflective learning) thissentence may be represented as an axiom and put into the appropriate separate micro-theory as the following CycL fragment
($forAll X ($implies
($isa X $LearningObject)
($resourceType X $Active))
Then resources that are not active are simply excluded from the notion itself This is astrong position in the sense that precludes some views on the learning design process
Table 2 Some examples of theoretical assumptions and guidelines
Archetypicalstandpoint Assumptions Guidelines
Instructivist(Behaviouristic)
bull Learning occurs through association and reinforcement
bull The results of learning can be appreciated through observable outcomes
(GB1) Every instance of oc-Learning entails theexposure to some LearningObjectthat produces the association (GB2) Learning Activitysequences should provide assessmentactivities that confirm theassociations that should have beenestablished so far
Constructivist bull Learning proceeds by building mental structures
(GC1) Agentrsquos mental structuresshould be represented(GC2) These structures should bepre-assessed in order to becomparable with the resulting ones(GC3) Design Activities arerecommended(GC4) Contrast Activities arerecommended
Sociocultural bull Learning occurs through social participatio
(GSC1) Every instance ofoc-Learning entails theinteraction of severaloc-IntelligentAgents asoc-Actors
344 British Journal of Educational Technology Vol 37 No 3 2006
copy 2006 The Author Journal compilation copy 2006 British Educational Communications and Technology Agency
Another typical example may be that of a sociocultural guideline that preclude Methodsthat do not entail multirole activities that provide access to social interaction ServicesThis approach to guidelines provides the best framework for automation and wouldenable the creation of microtheories that represent clearly formulated views onlearning Nonetheless there are two major problems for the use of this option as theunique mechanism for encoding design rationales
1 There is no clear-cut distinction between models and theories of learning becausethey come from a history of schools of thought with interrelations (Palmer 2001)This would lead to the need for the representation of dozens of variants reflectingpositions that only diverge in some aspects
2 Many guidelines cannot be represented as logical sentences because theyrepresent lsquorecommendationsrsquo or vague requirements eg see (GC3) or (GC4)(Table 2)
These issues suggest that a combination of formal sentences with reified guidelinesprovide the best option for flexibility The representation of guidelines has beenstudied in the field of HumanndashComputer-Interaction (HCI) in the last decade(Vanderdonckt 1999) and the main philosophy of such representations remainsvalid in the learning design domain The major challenge for such representations isthat of providing computational semantics that could be used for the implementationof software support but this is difficult to attain in all the cases Thus the concept oflearning design guideline will provide two kinds of representations general andschema-specific
The general use of instances of oc_NormativeSpecification allows the model-ling of lsquoconceptual works that describe how something should bersquo This may includerepresentation of known normative representations or procedures For example onemay refer to a concrete learning design model as that derived from Bannister (1981)which states that creativity cycles follow an initial phase of brainstorming (idea cap-ture) followed by a phase in which key issues are focused into an operation strategy(idea development) and a third phase in which project control takes place through alsquorecursive cyclersquo (operational management) The reference could be made by pointing tothe academic reference describing it as a conceptual work or it could be described asan Activity-Spec modelled explicitly inside the ontology As another examplesome detailed guidelines could be modelled by the normative specification sub-category of oc-Instructions defined as lsquoa sequence of tasks to be performed byan agentrsquo
Schema-specific guidelines restrict the freedom to represent criteria as conceptual workinstances to a concrete and commonly agreed ontology that determine what can be saidabout how learning designs should be according to a theoretical standpoint or frame-work These build on an activityndashrole ontology as the one described in the previoussection and add a prescription on the kind of lsquoshould-bersquo predications that are allow-able Table 3 describes examples of a minimal language that could be used for thatpurpose
Semantic learning designs 345
copy 2006 The Author Journal compilation copy 2006 British Educational Communications and Technology Agency
In the examples CAT represents a category of individual expressed through a conceptor a logical formula and the sentence is a synthetic way of expressing some recom-mended or mandatory constraint over such category Other languages could be engi-neered in further research but the one described can be directly expressed in OpenCycand combined with software modules that check or enforce them in some way Notethat this representation does not use direct OpenCyc sentences so that reification is usedinstead of direct logical sentences It is not the aim of this paper to determine the mostpowerful and flexible representation language but only set a foundation for furtherspecification efforts
Analysis and comparison of two theoretical semantic frameworksAs an assessment of the representational capabilities of the scheme just described acase study of representation is sketched in this section We will use the learning designtoolkit described by Conole et al (2004) as the underlying terminology even thoughother frameworks could be used instead
Two rather diverging positions for design will be used as an illustration Table 4 sum-marises the two design scenarios
Table 3 Possible language for use in learning designs
Sentence type Potential uses Example
RecommendationshouldBe (CATSENT)
Recommend kinds ofactivities
Design activities are recommended (GC3)shouldBe(Method subEvent (CATDesignActivity))
ObligationmustBe (CATSENT)
Some activity typesmust follow others
Assessments must follow each expositionto contents (GD1) mustBe(and(ExpositiveActivityIndividualActivity)startAfterEndingOf(CAT
AssessmentActivity)
Activities must havesome characteristic
Every activity must entail interactionmustBe(LDActivity(usedIn(CAT
InteractiveService)))Every activity must provide acontent-oriented learning object forindividual use (similar to GB1)mustBe(LDActivity(and (usedIn(CAT
LearningObject)(learnerRoleNumber(1))))
Some structure shouldbe enforced in methods
Methods should have a sequential actstructuremustBe(Methodnot ((subEvent(CAT Play))))
CAT = category GC3 = Constructivist Guideline 3 (see Table 2) GD1 = Guideline 1
346 British Journal of Educational Technology Vol 37 No 3 2006
copy 2006 The Author Journal compilation copy 2006 British Educational Communications and Technology Agency
The following subsections sketch the main outcomes of an example learning designprocess for both scenarios assuming that the process described by Conole et al (2004)is used in both situations
Differences in modelling the design processThe forces driving the design process in both scenarios are fairly diverging The designof Scenario A will typically be constrained by a clear target specified in terms of anaggregated competency level while in Scenario B there are no concrete competenciesto be acquired and the main objective of the design activity would be twofold On theone hand the activities are targeted at promoting interchange of ideas in an attemptto stimulate casual convergence of interest which is the main value of the culture ofthe special interest group On the other hand the activities have the generic objectiveof strengthening social ties which is an important element of cohesion for the under-taking of joint projects
Table 5 describes the main inputs and possible activities and their objectives at a highlevel
The changes intended to be achieved in each of the scenarios are stated differently witha formal definition in (A) and a shallow cognitive and social representation in the case
Table 4 Learning design scenarios
Scenario DescriptionClassification according to
Conole et al (2004) toolkit
(A) Systemic-Behaviourist A software development company attempts to designlearning units about newtechnologies or softwareproducts for its developerstaff The design is drivenby a cycle of lsquocompetencygap analysisrsquo that modelsorganisational learning in terms of the aggregated competencies of the employeesmdashsimilar to theone described by Sicilia(2005)
Focuses on the individual where a body of informationforms the basis and rawmaterial for learning Learning is mainly considered as nonreflective ie based on skill learning and memorisation
(B) Sociocultural A special interest group ona concrete research topicaims at developing the knowledge of its membersby sharing experiencesand discussing on themesof common interest
Focuses on the social whereexperience is the main source of learning Conscious reflection on experience is transformed into learning
Semantic learning designs 347
copy 2006 The Author Journal compilation copy 2006 British Educational Communications and Technology Agency
of (B) In addition case (B) requires a model of social network that is largely irrelevantin case (A)
Differences in recording activity rationalesThe approach for the recording of design rationales also differs because of the differentcontexts Table 6 summarises the main elements for each of the scenarios
Conclusions and directions for researchCurrent technology and practice related to the description of reusable digital learningresources has resulted in techniques that allow the description of learning activities of
Table 5 Inputs activities and objectives of two learning design scenarios
Scenario Design inputs Activities and their objectives
(A) Systemic-Behaviourist A formal competency recordfor employees and competencyrequirements expressed in terms of amounts of required competencies (assuming a shared measurement scale) Such requirements are actually derived from strategy or from the contingent requirements of upcoming projects
Competency instances are used to measure the levelof Competency-Spec for the required competencies (or their constituents) A formal match considering the agenda of employees is carried out for determining the target learners LearningObjects forindividual use are selected asthe main elements of theactivities and the criteria forselection are again a match ofthe objectives stated inmetadata (as stated incategory 9 of Learning Object MetadatamdashClassification) and thecompetencies required
(B) Sociocultural The expertise of the special interest group members as stated in the concepts addressed by their publicationsprojects or other activities
Informal group activities are the main approach applicablesince there is not a predefinedgroup structure but a cultureof sharing and socialisation The results of the sharing of ideas can be expressed in terms of oc-knowsAboutpredicates and social ties will be modelled through positive valence FOAF-knows(httpxmlnscomfoaf01)predicates with an added numerical value representingstrength
348 British Journal of Educational Technology Vol 37 No 3 2006
copy 2006 The Author Journal compilation copy 2006 British Educational Communications and Technology Agency
a complex kind including their intended participants educational objectives and exe-cution constraints The representation and recording of the assumptions and guidelinesthat were considered in the creation of such resources would extend current techniquesto capture the rationale of the design itself This in turn will eventually result in theavailability of a large base of empirical data available in which theoretical assumptionsor rules are explicitly connected to the resulting designs as these could be linked withthe actual results and trace of the learning processes that used them The resulting bodyof evidence would represent a critical asset in the development of the theory of learningmediated through computers
This paper has described the first attempt to extend existing learning technology modelswith a schema explicitly targeted at describing learning design rationales The OpenCycknowledge base has been used as the semantic reference from which the ontologicalschema has been specified Concretely an activity-based schema based on IMS LD hasbeen described and theoretical assumptions and codified guidelines have been intro-duced The flexibility of the representation schema has been analysed through therepresentation of two stereotyped theoretical positions
Two main research directions should continue the work described here On the onehand further ontological engineering effort is required to come up with a more com-prehensive and rich schema for learning design rationales On the other hand tool
Table 6 Main elements of the two learning design scenarios
Scenario Approach Examples
(A) Systemic-Behaviourist Formal logical sentencesare used to guarantee that every design followsan organisationalobjective
It is required that the objective of every Method is attached to a Competency-Spec instance as objective
Formal assessment is a requirement foraccountability purposes
Guideline (GD1) is attached to everyDesigningLearningResource instance that produces an Act
(B) Sociocultural The interaction in informal settings can beused to derive social models (Sicilia amp Garciacutea2004)
The interactions in discussion canbe used to measure the degree of lsquoproximityrsquo between members Therationale of the activities can bejustified in terms of SocialTies inside SocialNetworks Further the focus of each groupactivity should be clearly stated anda combination of the increase in collective knowledge and the strengthening of social relations can be used for assessment
Semantic learning designs 349
copy 2006 The Author Journal compilation copy 2006 British Educational Communications and Technology Agency
implementation and record gathering efforts would be required to attain the ultimateobjective of the ontological definitions described so far With such effort an unprece-dented research material could be available in some years providing improved oppor-tunities for scientific inquiry on design as an activity regarding pedagogy
Other potential directions for research may be the description of interpretations of thedesign intentions (for which no rationale was explicitly recorded) the mining or detec-tion of patterns as related to rationale description or the detailed use of theoreticalrationales in agents that support the learning process which represent a category ofrelevant learning support software (Holmes in press)
ReferencesAllert H (2004) Coherent social systems for learningmdashan approach for contextualized and
community-centred metadata Journal of Interactive Media in Education 2004 1Baader F Calvanese D McGuinness D Nardi D amp Patel-Schneider P (Eds) (2003) The
description logic handbook Theory implementation and applications Cambridge CambridgeUniversity Press
Bannister D (1981) Personal construct theory and research method In P Reason amp J Rowan(Eds) Human inquiry a sourcebook of new paradigm research (pp 191ndash200) Chichester UKJohn Wiley and Sons
Berners-Lee T Hendler J amp Lassila O (2001) The semantic web Scientific American 284 534ndash43
Brown A amp Voltz B (2005) Elements of effective e-learning design International Review ofResearch in Open and Distance Learning 6 1
Buzza D Bean D Harrigan K amp Carey T (2005) Learning design repositories adaptinglearning design specifications for shared instructional knowledge Canadian Journal of Learningand Technology 30 3
Conole G Dyke M Oliver M amp Seale J (2004) Mapping pedagogy and tools for effectivelearning design Computers amp Education 43 1ndash2 17ndash33
Dodero J M Aedo I amp Diacuteaz P (2002) Participative knowledge production of learning objectsfor e-books The Electronic Library 20 4 296ndash305
Friesen N (2005) Interoperability and learning objects an overview of e-learning standardiza-tion Interdisciplinary Journal of Knowledge and Learning Objects 1 1 23ndash31
Gruber T (1995) Towards principles for the design of ontologies used for knowledge sharingInternational Journal of HumanndashComputer Studies 43 56 907ndash928
Hatchuel A (2002) Towards design theory and expandable rationality the unfinished programof Herbert Simon Journal of Management and Governance 5 3ndash4 260ndash271
Holmes J (in press) Designing agents to support learning by explaining Computers and EducationIEEE Learning Technology Standards Committee (2002) Learning object metadata (LOM) Final
Draft Standard IEEE 1484121ndash2002IMS (2003) IMS Learning Design Information Model (Version 10) Final Specification 20
January 2003Kaptelinin V Kuutti K amp Bannon L (1995) Activity theory basic concepts and applications
In Blumenthalet al (Eds) Humanndashcomputer interaction Lecture Notes in Computer Science 1015189ndash201
Koper R (2004) Use of the semantic web to solve some basic problems in education increaseflexible distributed lifelong learning decrease teacherrsquos workload Journal of Interactive Mediain Education 2004 6
Lenat D (1995) CYC a large-scale investment in knowledge infrastructure Communications ofthe ACM 38 11 33ndash38
350 British Journal of Educational Technology Vol 37 No 3 2006
copy 2006 The Author Journal compilation copy 2006 British Educational Communications and Technology Agency
McGreal R (2004) Learning objects a practical definition International Journal of InstructionalTechnology and Distance Learning 1 9 21ndash32
Packer M amp Goicoechea J (2000) Sociocultural and constructivist theories of learningontology not just epistemology Educational Psychologist 35 4 227ndash241
Palmer J (2001) Fifty major thinkers on education London and New York RoutledgePolsani P R (2003) Use and abuse of reusable learning objects Journal of Digital Information 3
4 Rothwell W amp Kazanas H (1992) Mastering the instructional design process San Francisco CA
Jossey-BassSaacutenchez-Alonso S amp Sicilia M A (2004) Relationships and commitments in learning object
metadata In Proceedings of the 5th International Conference on Information Technology BasedHigher Education and Training ITHET 2004 Istanbul Turkey
Sicilia M A (2005) Ontology-Based Competency Management Infrastructures for theKnowledge-intensive Learning Organization In Lytras amp Naeve (Eds) Intelligent Learning Infra-structures in Knowledge Intensive Organizations A Semantic Web perspective IDEA USA pp 302ndash324
Sicilia M A amp Garciacutea E (2004) Fuzzy group models for adaptation in cooperative informationretrieval contexts Lecture Notes in Computer Science 2932 324ndash334
Sicilia M A amp Garciacutea E (2005) On the convergence of formal ontologies and standardizede-learning Journal of Distance Education Technologies 3 2 13ndash29
Sicilia M A amp Lytras M (2005) On the representation of change according to different ontol-ogies of learning International Journal of Learning and Change 1 1 66ndash79
Sicilia M A Lytras M Rodriacuteguez E amp Garciacutea E (in press) Integrating descriptions of knowl-edge management learning activities into large ontological structures a case study Data andKnowledge Engineering
Tennant M (1991) Expertise as a dimension of adult development New Education 13 1 49ndash55Vanderdonckt J (1999) Development milestones towards a tool for working with guidelines
Interacting with Computers 12 2 81ndash118Wand Y amp Weber R (1995) On the deep structure of information systems Information Systems
Journal 5 203ndash223Wiley D A (2001) The instructional use of learning objects Bloomington IN Association for
Educational Communications and Technology
344 British Journal of Educational Technology Vol 37 No 3 2006
copy 2006 The Author Journal compilation copy 2006 British Educational Communications and Technology Agency
Another typical example may be that of a sociocultural guideline that preclude Methodsthat do not entail multirole activities that provide access to social interaction ServicesThis approach to guidelines provides the best framework for automation and wouldenable the creation of microtheories that represent clearly formulated views onlearning Nonetheless there are two major problems for the use of this option as theunique mechanism for encoding design rationales
1 There is no clear-cut distinction between models and theories of learning becausethey come from a history of schools of thought with interrelations (Palmer 2001)This would lead to the need for the representation of dozens of variants reflectingpositions that only diverge in some aspects
2 Many guidelines cannot be represented as logical sentences because theyrepresent lsquorecommendationsrsquo or vague requirements eg see (GC3) or (GC4)(Table 2)
These issues suggest that a combination of formal sentences with reified guidelinesprovide the best option for flexibility The representation of guidelines has beenstudied in the field of HumanndashComputer-Interaction (HCI) in the last decade(Vanderdonckt 1999) and the main philosophy of such representations remainsvalid in the learning design domain The major challenge for such representations isthat of providing computational semantics that could be used for the implementationof software support but this is difficult to attain in all the cases Thus the concept oflearning design guideline will provide two kinds of representations general andschema-specific
The general use of instances of oc_NormativeSpecification allows the model-ling of lsquoconceptual works that describe how something should bersquo This may includerepresentation of known normative representations or procedures For example onemay refer to a concrete learning design model as that derived from Bannister (1981)which states that creativity cycles follow an initial phase of brainstorming (idea cap-ture) followed by a phase in which key issues are focused into an operation strategy(idea development) and a third phase in which project control takes place through alsquorecursive cyclersquo (operational management) The reference could be made by pointing tothe academic reference describing it as a conceptual work or it could be described asan Activity-Spec modelled explicitly inside the ontology As another examplesome detailed guidelines could be modelled by the normative specification sub-category of oc-Instructions defined as lsquoa sequence of tasks to be performed byan agentrsquo
Schema-specific guidelines restrict the freedom to represent criteria as conceptual workinstances to a concrete and commonly agreed ontology that determine what can be saidabout how learning designs should be according to a theoretical standpoint or frame-work These build on an activityndashrole ontology as the one described in the previoussection and add a prescription on the kind of lsquoshould-bersquo predications that are allow-able Table 3 describes examples of a minimal language that could be used for thatpurpose
Semantic learning designs 345
copy 2006 The Author Journal compilation copy 2006 British Educational Communications and Technology Agency
In the examples CAT represents a category of individual expressed through a conceptor a logical formula and the sentence is a synthetic way of expressing some recom-mended or mandatory constraint over such category Other languages could be engi-neered in further research but the one described can be directly expressed in OpenCycand combined with software modules that check or enforce them in some way Notethat this representation does not use direct OpenCyc sentences so that reification is usedinstead of direct logical sentences It is not the aim of this paper to determine the mostpowerful and flexible representation language but only set a foundation for furtherspecification efforts
Analysis and comparison of two theoretical semantic frameworksAs an assessment of the representational capabilities of the scheme just described acase study of representation is sketched in this section We will use the learning designtoolkit described by Conole et al (2004) as the underlying terminology even thoughother frameworks could be used instead
Two rather diverging positions for design will be used as an illustration Table 4 sum-marises the two design scenarios
Table 3 Possible language for use in learning designs
Sentence type Potential uses Example
RecommendationshouldBe (CATSENT)
Recommend kinds ofactivities
Design activities are recommended (GC3)shouldBe(Method subEvent (CATDesignActivity))
ObligationmustBe (CATSENT)
Some activity typesmust follow others
Assessments must follow each expositionto contents (GD1) mustBe(and(ExpositiveActivityIndividualActivity)startAfterEndingOf(CAT
AssessmentActivity)
Activities must havesome characteristic
Every activity must entail interactionmustBe(LDActivity(usedIn(CAT
InteractiveService)))Every activity must provide acontent-oriented learning object forindividual use (similar to GB1)mustBe(LDActivity(and (usedIn(CAT
LearningObject)(learnerRoleNumber(1))))
Some structure shouldbe enforced in methods
Methods should have a sequential actstructuremustBe(Methodnot ((subEvent(CAT Play))))
CAT = category GC3 = Constructivist Guideline 3 (see Table 2) GD1 = Guideline 1
346 British Journal of Educational Technology Vol 37 No 3 2006
copy 2006 The Author Journal compilation copy 2006 British Educational Communications and Technology Agency
The following subsections sketch the main outcomes of an example learning designprocess for both scenarios assuming that the process described by Conole et al (2004)is used in both situations
Differences in modelling the design processThe forces driving the design process in both scenarios are fairly diverging The designof Scenario A will typically be constrained by a clear target specified in terms of anaggregated competency level while in Scenario B there are no concrete competenciesto be acquired and the main objective of the design activity would be twofold On theone hand the activities are targeted at promoting interchange of ideas in an attemptto stimulate casual convergence of interest which is the main value of the culture ofthe special interest group On the other hand the activities have the generic objectiveof strengthening social ties which is an important element of cohesion for the under-taking of joint projects
Table 5 describes the main inputs and possible activities and their objectives at a highlevel
The changes intended to be achieved in each of the scenarios are stated differently witha formal definition in (A) and a shallow cognitive and social representation in the case
Table 4 Learning design scenarios
Scenario DescriptionClassification according to
Conole et al (2004) toolkit
(A) Systemic-Behaviourist A software development company attempts to designlearning units about newtechnologies or softwareproducts for its developerstaff The design is drivenby a cycle of lsquocompetencygap analysisrsquo that modelsorganisational learning in terms of the aggregated competencies of the employeesmdashsimilar to theone described by Sicilia(2005)
Focuses on the individual where a body of informationforms the basis and rawmaterial for learning Learning is mainly considered as nonreflective ie based on skill learning and memorisation
(B) Sociocultural A special interest group ona concrete research topicaims at developing the knowledge of its membersby sharing experiencesand discussing on themesof common interest
Focuses on the social whereexperience is the main source of learning Conscious reflection on experience is transformed into learning
Semantic learning designs 347
copy 2006 The Author Journal compilation copy 2006 British Educational Communications and Technology Agency
of (B) In addition case (B) requires a model of social network that is largely irrelevantin case (A)
Differences in recording activity rationalesThe approach for the recording of design rationales also differs because of the differentcontexts Table 6 summarises the main elements for each of the scenarios
Conclusions and directions for researchCurrent technology and practice related to the description of reusable digital learningresources has resulted in techniques that allow the description of learning activities of
Table 5 Inputs activities and objectives of two learning design scenarios
Scenario Design inputs Activities and their objectives
(A) Systemic-Behaviourist A formal competency recordfor employees and competencyrequirements expressed in terms of amounts of required competencies (assuming a shared measurement scale) Such requirements are actually derived from strategy or from the contingent requirements of upcoming projects
Competency instances are used to measure the levelof Competency-Spec for the required competencies (or their constituents) A formal match considering the agenda of employees is carried out for determining the target learners LearningObjects forindividual use are selected asthe main elements of theactivities and the criteria forselection are again a match ofthe objectives stated inmetadata (as stated incategory 9 of Learning Object MetadatamdashClassification) and thecompetencies required
(B) Sociocultural The expertise of the special interest group members as stated in the concepts addressed by their publicationsprojects or other activities
Informal group activities are the main approach applicablesince there is not a predefinedgroup structure but a cultureof sharing and socialisation The results of the sharing of ideas can be expressed in terms of oc-knowsAboutpredicates and social ties will be modelled through positive valence FOAF-knows(httpxmlnscomfoaf01)predicates with an added numerical value representingstrength
348 British Journal of Educational Technology Vol 37 No 3 2006
copy 2006 The Author Journal compilation copy 2006 British Educational Communications and Technology Agency
a complex kind including their intended participants educational objectives and exe-cution constraints The representation and recording of the assumptions and guidelinesthat were considered in the creation of such resources would extend current techniquesto capture the rationale of the design itself This in turn will eventually result in theavailability of a large base of empirical data available in which theoretical assumptionsor rules are explicitly connected to the resulting designs as these could be linked withthe actual results and trace of the learning processes that used them The resulting bodyof evidence would represent a critical asset in the development of the theory of learningmediated through computers
This paper has described the first attempt to extend existing learning technology modelswith a schema explicitly targeted at describing learning design rationales The OpenCycknowledge base has been used as the semantic reference from which the ontologicalschema has been specified Concretely an activity-based schema based on IMS LD hasbeen described and theoretical assumptions and codified guidelines have been intro-duced The flexibility of the representation schema has been analysed through therepresentation of two stereotyped theoretical positions
Two main research directions should continue the work described here On the onehand further ontological engineering effort is required to come up with a more com-prehensive and rich schema for learning design rationales On the other hand tool
Table 6 Main elements of the two learning design scenarios
Scenario Approach Examples
(A) Systemic-Behaviourist Formal logical sentencesare used to guarantee that every design followsan organisationalobjective
It is required that the objective of every Method is attached to a Competency-Spec instance as objective
Formal assessment is a requirement foraccountability purposes
Guideline (GD1) is attached to everyDesigningLearningResource instance that produces an Act
(B) Sociocultural The interaction in informal settings can beused to derive social models (Sicilia amp Garciacutea2004)
The interactions in discussion canbe used to measure the degree of lsquoproximityrsquo between members Therationale of the activities can bejustified in terms of SocialTies inside SocialNetworks Further the focus of each groupactivity should be clearly stated anda combination of the increase in collective knowledge and the strengthening of social relations can be used for assessment
Semantic learning designs 349
copy 2006 The Author Journal compilation copy 2006 British Educational Communications and Technology Agency
implementation and record gathering efforts would be required to attain the ultimateobjective of the ontological definitions described so far With such effort an unprece-dented research material could be available in some years providing improved oppor-tunities for scientific inquiry on design as an activity regarding pedagogy
Other potential directions for research may be the description of interpretations of thedesign intentions (for which no rationale was explicitly recorded) the mining or detec-tion of patterns as related to rationale description or the detailed use of theoreticalrationales in agents that support the learning process which represent a category ofrelevant learning support software (Holmes in press)
ReferencesAllert H (2004) Coherent social systems for learningmdashan approach for contextualized and
community-centred metadata Journal of Interactive Media in Education 2004 1Baader F Calvanese D McGuinness D Nardi D amp Patel-Schneider P (Eds) (2003) The
description logic handbook Theory implementation and applications Cambridge CambridgeUniversity Press
Bannister D (1981) Personal construct theory and research method In P Reason amp J Rowan(Eds) Human inquiry a sourcebook of new paradigm research (pp 191ndash200) Chichester UKJohn Wiley and Sons
Berners-Lee T Hendler J amp Lassila O (2001) The semantic web Scientific American 284 534ndash43
Brown A amp Voltz B (2005) Elements of effective e-learning design International Review ofResearch in Open and Distance Learning 6 1
Buzza D Bean D Harrigan K amp Carey T (2005) Learning design repositories adaptinglearning design specifications for shared instructional knowledge Canadian Journal of Learningand Technology 30 3
Conole G Dyke M Oliver M amp Seale J (2004) Mapping pedagogy and tools for effectivelearning design Computers amp Education 43 1ndash2 17ndash33
Dodero J M Aedo I amp Diacuteaz P (2002) Participative knowledge production of learning objectsfor e-books The Electronic Library 20 4 296ndash305
Friesen N (2005) Interoperability and learning objects an overview of e-learning standardiza-tion Interdisciplinary Journal of Knowledge and Learning Objects 1 1 23ndash31
Gruber T (1995) Towards principles for the design of ontologies used for knowledge sharingInternational Journal of HumanndashComputer Studies 43 56 907ndash928
Hatchuel A (2002) Towards design theory and expandable rationality the unfinished programof Herbert Simon Journal of Management and Governance 5 3ndash4 260ndash271
Holmes J (in press) Designing agents to support learning by explaining Computers and EducationIEEE Learning Technology Standards Committee (2002) Learning object metadata (LOM) Final
Draft Standard IEEE 1484121ndash2002IMS (2003) IMS Learning Design Information Model (Version 10) Final Specification 20
January 2003Kaptelinin V Kuutti K amp Bannon L (1995) Activity theory basic concepts and applications
In Blumenthalet al (Eds) Humanndashcomputer interaction Lecture Notes in Computer Science 1015189ndash201
Koper R (2004) Use of the semantic web to solve some basic problems in education increaseflexible distributed lifelong learning decrease teacherrsquos workload Journal of Interactive Mediain Education 2004 6
Lenat D (1995) CYC a large-scale investment in knowledge infrastructure Communications ofthe ACM 38 11 33ndash38
350 British Journal of Educational Technology Vol 37 No 3 2006
copy 2006 The Author Journal compilation copy 2006 British Educational Communications and Technology Agency
McGreal R (2004) Learning objects a practical definition International Journal of InstructionalTechnology and Distance Learning 1 9 21ndash32
Packer M amp Goicoechea J (2000) Sociocultural and constructivist theories of learningontology not just epistemology Educational Psychologist 35 4 227ndash241
Palmer J (2001) Fifty major thinkers on education London and New York RoutledgePolsani P R (2003) Use and abuse of reusable learning objects Journal of Digital Information 3
4 Rothwell W amp Kazanas H (1992) Mastering the instructional design process San Francisco CA
Jossey-BassSaacutenchez-Alonso S amp Sicilia M A (2004) Relationships and commitments in learning object
metadata In Proceedings of the 5th International Conference on Information Technology BasedHigher Education and Training ITHET 2004 Istanbul Turkey
Sicilia M A (2005) Ontology-Based Competency Management Infrastructures for theKnowledge-intensive Learning Organization In Lytras amp Naeve (Eds) Intelligent Learning Infra-structures in Knowledge Intensive Organizations A Semantic Web perspective IDEA USA pp 302ndash324
Sicilia M A amp Garciacutea E (2004) Fuzzy group models for adaptation in cooperative informationretrieval contexts Lecture Notes in Computer Science 2932 324ndash334
Sicilia M A amp Garciacutea E (2005) On the convergence of formal ontologies and standardizede-learning Journal of Distance Education Technologies 3 2 13ndash29
Sicilia M A amp Lytras M (2005) On the representation of change according to different ontol-ogies of learning International Journal of Learning and Change 1 1 66ndash79
Sicilia M A Lytras M Rodriacuteguez E amp Garciacutea E (in press) Integrating descriptions of knowl-edge management learning activities into large ontological structures a case study Data andKnowledge Engineering
Tennant M (1991) Expertise as a dimension of adult development New Education 13 1 49ndash55Vanderdonckt J (1999) Development milestones towards a tool for working with guidelines
Interacting with Computers 12 2 81ndash118Wand Y amp Weber R (1995) On the deep structure of information systems Information Systems
Journal 5 203ndash223Wiley D A (2001) The instructional use of learning objects Bloomington IN Association for
Educational Communications and Technology
Semantic learning designs 345
copy 2006 The Author Journal compilation copy 2006 British Educational Communications and Technology Agency
In the examples CAT represents a category of individual expressed through a conceptor a logical formula and the sentence is a synthetic way of expressing some recom-mended or mandatory constraint over such category Other languages could be engi-neered in further research but the one described can be directly expressed in OpenCycand combined with software modules that check or enforce them in some way Notethat this representation does not use direct OpenCyc sentences so that reification is usedinstead of direct logical sentences It is not the aim of this paper to determine the mostpowerful and flexible representation language but only set a foundation for furtherspecification efforts
Analysis and comparison of two theoretical semantic frameworksAs an assessment of the representational capabilities of the scheme just described acase study of representation is sketched in this section We will use the learning designtoolkit described by Conole et al (2004) as the underlying terminology even thoughother frameworks could be used instead
Two rather diverging positions for design will be used as an illustration Table 4 sum-marises the two design scenarios
Table 3 Possible language for use in learning designs
Sentence type Potential uses Example
RecommendationshouldBe (CATSENT)
Recommend kinds ofactivities
Design activities are recommended (GC3)shouldBe(Method subEvent (CATDesignActivity))
ObligationmustBe (CATSENT)
Some activity typesmust follow others
Assessments must follow each expositionto contents (GD1) mustBe(and(ExpositiveActivityIndividualActivity)startAfterEndingOf(CAT
AssessmentActivity)
Activities must havesome characteristic
Every activity must entail interactionmustBe(LDActivity(usedIn(CAT
InteractiveService)))Every activity must provide acontent-oriented learning object forindividual use (similar to GB1)mustBe(LDActivity(and (usedIn(CAT
LearningObject)(learnerRoleNumber(1))))
Some structure shouldbe enforced in methods
Methods should have a sequential actstructuremustBe(Methodnot ((subEvent(CAT Play))))
CAT = category GC3 = Constructivist Guideline 3 (see Table 2) GD1 = Guideline 1
346 British Journal of Educational Technology Vol 37 No 3 2006
copy 2006 The Author Journal compilation copy 2006 British Educational Communications and Technology Agency
The following subsections sketch the main outcomes of an example learning designprocess for both scenarios assuming that the process described by Conole et al (2004)is used in both situations
Differences in modelling the design processThe forces driving the design process in both scenarios are fairly diverging The designof Scenario A will typically be constrained by a clear target specified in terms of anaggregated competency level while in Scenario B there are no concrete competenciesto be acquired and the main objective of the design activity would be twofold On theone hand the activities are targeted at promoting interchange of ideas in an attemptto stimulate casual convergence of interest which is the main value of the culture ofthe special interest group On the other hand the activities have the generic objectiveof strengthening social ties which is an important element of cohesion for the under-taking of joint projects
Table 5 describes the main inputs and possible activities and their objectives at a highlevel
The changes intended to be achieved in each of the scenarios are stated differently witha formal definition in (A) and a shallow cognitive and social representation in the case
Table 4 Learning design scenarios
Scenario DescriptionClassification according to
Conole et al (2004) toolkit
(A) Systemic-Behaviourist A software development company attempts to designlearning units about newtechnologies or softwareproducts for its developerstaff The design is drivenby a cycle of lsquocompetencygap analysisrsquo that modelsorganisational learning in terms of the aggregated competencies of the employeesmdashsimilar to theone described by Sicilia(2005)
Focuses on the individual where a body of informationforms the basis and rawmaterial for learning Learning is mainly considered as nonreflective ie based on skill learning and memorisation
(B) Sociocultural A special interest group ona concrete research topicaims at developing the knowledge of its membersby sharing experiencesand discussing on themesof common interest
Focuses on the social whereexperience is the main source of learning Conscious reflection on experience is transformed into learning
Semantic learning designs 347
copy 2006 The Author Journal compilation copy 2006 British Educational Communications and Technology Agency
of (B) In addition case (B) requires a model of social network that is largely irrelevantin case (A)
Differences in recording activity rationalesThe approach for the recording of design rationales also differs because of the differentcontexts Table 6 summarises the main elements for each of the scenarios
Conclusions and directions for researchCurrent technology and practice related to the description of reusable digital learningresources has resulted in techniques that allow the description of learning activities of
Table 5 Inputs activities and objectives of two learning design scenarios
Scenario Design inputs Activities and their objectives
(A) Systemic-Behaviourist A formal competency recordfor employees and competencyrequirements expressed in terms of amounts of required competencies (assuming a shared measurement scale) Such requirements are actually derived from strategy or from the contingent requirements of upcoming projects
Competency instances are used to measure the levelof Competency-Spec for the required competencies (or their constituents) A formal match considering the agenda of employees is carried out for determining the target learners LearningObjects forindividual use are selected asthe main elements of theactivities and the criteria forselection are again a match ofthe objectives stated inmetadata (as stated incategory 9 of Learning Object MetadatamdashClassification) and thecompetencies required
(B) Sociocultural The expertise of the special interest group members as stated in the concepts addressed by their publicationsprojects or other activities
Informal group activities are the main approach applicablesince there is not a predefinedgroup structure but a cultureof sharing and socialisation The results of the sharing of ideas can be expressed in terms of oc-knowsAboutpredicates and social ties will be modelled through positive valence FOAF-knows(httpxmlnscomfoaf01)predicates with an added numerical value representingstrength
348 British Journal of Educational Technology Vol 37 No 3 2006
copy 2006 The Author Journal compilation copy 2006 British Educational Communications and Technology Agency
a complex kind including their intended participants educational objectives and exe-cution constraints The representation and recording of the assumptions and guidelinesthat were considered in the creation of such resources would extend current techniquesto capture the rationale of the design itself This in turn will eventually result in theavailability of a large base of empirical data available in which theoretical assumptionsor rules are explicitly connected to the resulting designs as these could be linked withthe actual results and trace of the learning processes that used them The resulting bodyof evidence would represent a critical asset in the development of the theory of learningmediated through computers
This paper has described the first attempt to extend existing learning technology modelswith a schema explicitly targeted at describing learning design rationales The OpenCycknowledge base has been used as the semantic reference from which the ontologicalschema has been specified Concretely an activity-based schema based on IMS LD hasbeen described and theoretical assumptions and codified guidelines have been intro-duced The flexibility of the representation schema has been analysed through therepresentation of two stereotyped theoretical positions
Two main research directions should continue the work described here On the onehand further ontological engineering effort is required to come up with a more com-prehensive and rich schema for learning design rationales On the other hand tool
Table 6 Main elements of the two learning design scenarios
Scenario Approach Examples
(A) Systemic-Behaviourist Formal logical sentencesare used to guarantee that every design followsan organisationalobjective
It is required that the objective of every Method is attached to a Competency-Spec instance as objective
Formal assessment is a requirement foraccountability purposes
Guideline (GD1) is attached to everyDesigningLearningResource instance that produces an Act
(B) Sociocultural The interaction in informal settings can beused to derive social models (Sicilia amp Garciacutea2004)
The interactions in discussion canbe used to measure the degree of lsquoproximityrsquo between members Therationale of the activities can bejustified in terms of SocialTies inside SocialNetworks Further the focus of each groupactivity should be clearly stated anda combination of the increase in collective knowledge and the strengthening of social relations can be used for assessment
Semantic learning designs 349
copy 2006 The Author Journal compilation copy 2006 British Educational Communications and Technology Agency
implementation and record gathering efforts would be required to attain the ultimateobjective of the ontological definitions described so far With such effort an unprece-dented research material could be available in some years providing improved oppor-tunities for scientific inquiry on design as an activity regarding pedagogy
Other potential directions for research may be the description of interpretations of thedesign intentions (for which no rationale was explicitly recorded) the mining or detec-tion of patterns as related to rationale description or the detailed use of theoreticalrationales in agents that support the learning process which represent a category ofrelevant learning support software (Holmes in press)
ReferencesAllert H (2004) Coherent social systems for learningmdashan approach for contextualized and
community-centred metadata Journal of Interactive Media in Education 2004 1Baader F Calvanese D McGuinness D Nardi D amp Patel-Schneider P (Eds) (2003) The
description logic handbook Theory implementation and applications Cambridge CambridgeUniversity Press
Bannister D (1981) Personal construct theory and research method In P Reason amp J Rowan(Eds) Human inquiry a sourcebook of new paradigm research (pp 191ndash200) Chichester UKJohn Wiley and Sons
Berners-Lee T Hendler J amp Lassila O (2001) The semantic web Scientific American 284 534ndash43
Brown A amp Voltz B (2005) Elements of effective e-learning design International Review ofResearch in Open and Distance Learning 6 1
Buzza D Bean D Harrigan K amp Carey T (2005) Learning design repositories adaptinglearning design specifications for shared instructional knowledge Canadian Journal of Learningand Technology 30 3
Conole G Dyke M Oliver M amp Seale J (2004) Mapping pedagogy and tools for effectivelearning design Computers amp Education 43 1ndash2 17ndash33
Dodero J M Aedo I amp Diacuteaz P (2002) Participative knowledge production of learning objectsfor e-books The Electronic Library 20 4 296ndash305
Friesen N (2005) Interoperability and learning objects an overview of e-learning standardiza-tion Interdisciplinary Journal of Knowledge and Learning Objects 1 1 23ndash31
Gruber T (1995) Towards principles for the design of ontologies used for knowledge sharingInternational Journal of HumanndashComputer Studies 43 56 907ndash928
Hatchuel A (2002) Towards design theory and expandable rationality the unfinished programof Herbert Simon Journal of Management and Governance 5 3ndash4 260ndash271
Holmes J (in press) Designing agents to support learning by explaining Computers and EducationIEEE Learning Technology Standards Committee (2002) Learning object metadata (LOM) Final
Draft Standard IEEE 1484121ndash2002IMS (2003) IMS Learning Design Information Model (Version 10) Final Specification 20
January 2003Kaptelinin V Kuutti K amp Bannon L (1995) Activity theory basic concepts and applications
In Blumenthalet al (Eds) Humanndashcomputer interaction Lecture Notes in Computer Science 1015189ndash201
Koper R (2004) Use of the semantic web to solve some basic problems in education increaseflexible distributed lifelong learning decrease teacherrsquos workload Journal of Interactive Mediain Education 2004 6
Lenat D (1995) CYC a large-scale investment in knowledge infrastructure Communications ofthe ACM 38 11 33ndash38
350 British Journal of Educational Technology Vol 37 No 3 2006
copy 2006 The Author Journal compilation copy 2006 British Educational Communications and Technology Agency
McGreal R (2004) Learning objects a practical definition International Journal of InstructionalTechnology and Distance Learning 1 9 21ndash32
Packer M amp Goicoechea J (2000) Sociocultural and constructivist theories of learningontology not just epistemology Educational Psychologist 35 4 227ndash241
Palmer J (2001) Fifty major thinkers on education London and New York RoutledgePolsani P R (2003) Use and abuse of reusable learning objects Journal of Digital Information 3
4 Rothwell W amp Kazanas H (1992) Mastering the instructional design process San Francisco CA
Jossey-BassSaacutenchez-Alonso S amp Sicilia M A (2004) Relationships and commitments in learning object
metadata In Proceedings of the 5th International Conference on Information Technology BasedHigher Education and Training ITHET 2004 Istanbul Turkey
Sicilia M A (2005) Ontology-Based Competency Management Infrastructures for theKnowledge-intensive Learning Organization In Lytras amp Naeve (Eds) Intelligent Learning Infra-structures in Knowledge Intensive Organizations A Semantic Web perspective IDEA USA pp 302ndash324
Sicilia M A amp Garciacutea E (2004) Fuzzy group models for adaptation in cooperative informationretrieval contexts Lecture Notes in Computer Science 2932 324ndash334
Sicilia M A amp Garciacutea E (2005) On the convergence of formal ontologies and standardizede-learning Journal of Distance Education Technologies 3 2 13ndash29
Sicilia M A amp Lytras M (2005) On the representation of change according to different ontol-ogies of learning International Journal of Learning and Change 1 1 66ndash79
Sicilia M A Lytras M Rodriacuteguez E amp Garciacutea E (in press) Integrating descriptions of knowl-edge management learning activities into large ontological structures a case study Data andKnowledge Engineering
Tennant M (1991) Expertise as a dimension of adult development New Education 13 1 49ndash55Vanderdonckt J (1999) Development milestones towards a tool for working with guidelines
Interacting with Computers 12 2 81ndash118Wand Y amp Weber R (1995) On the deep structure of information systems Information Systems
Journal 5 203ndash223Wiley D A (2001) The instructional use of learning objects Bloomington IN Association for
Educational Communications and Technology
346 British Journal of Educational Technology Vol 37 No 3 2006
copy 2006 The Author Journal compilation copy 2006 British Educational Communications and Technology Agency
The following subsections sketch the main outcomes of an example learning designprocess for both scenarios assuming that the process described by Conole et al (2004)is used in both situations
Differences in modelling the design processThe forces driving the design process in both scenarios are fairly diverging The designof Scenario A will typically be constrained by a clear target specified in terms of anaggregated competency level while in Scenario B there are no concrete competenciesto be acquired and the main objective of the design activity would be twofold On theone hand the activities are targeted at promoting interchange of ideas in an attemptto stimulate casual convergence of interest which is the main value of the culture ofthe special interest group On the other hand the activities have the generic objectiveof strengthening social ties which is an important element of cohesion for the under-taking of joint projects
Table 5 describes the main inputs and possible activities and their objectives at a highlevel
The changes intended to be achieved in each of the scenarios are stated differently witha formal definition in (A) and a shallow cognitive and social representation in the case
Table 4 Learning design scenarios
Scenario DescriptionClassification according to
Conole et al (2004) toolkit
(A) Systemic-Behaviourist A software development company attempts to designlearning units about newtechnologies or softwareproducts for its developerstaff The design is drivenby a cycle of lsquocompetencygap analysisrsquo that modelsorganisational learning in terms of the aggregated competencies of the employeesmdashsimilar to theone described by Sicilia(2005)
Focuses on the individual where a body of informationforms the basis and rawmaterial for learning Learning is mainly considered as nonreflective ie based on skill learning and memorisation
(B) Sociocultural A special interest group ona concrete research topicaims at developing the knowledge of its membersby sharing experiencesand discussing on themesof common interest
Focuses on the social whereexperience is the main source of learning Conscious reflection on experience is transformed into learning
Semantic learning designs 347
copy 2006 The Author Journal compilation copy 2006 British Educational Communications and Technology Agency
of (B) In addition case (B) requires a model of social network that is largely irrelevantin case (A)
Differences in recording activity rationalesThe approach for the recording of design rationales also differs because of the differentcontexts Table 6 summarises the main elements for each of the scenarios
Conclusions and directions for researchCurrent technology and practice related to the description of reusable digital learningresources has resulted in techniques that allow the description of learning activities of
Table 5 Inputs activities and objectives of two learning design scenarios
Scenario Design inputs Activities and their objectives
(A) Systemic-Behaviourist A formal competency recordfor employees and competencyrequirements expressed in terms of amounts of required competencies (assuming a shared measurement scale) Such requirements are actually derived from strategy or from the contingent requirements of upcoming projects
Competency instances are used to measure the levelof Competency-Spec for the required competencies (or their constituents) A formal match considering the agenda of employees is carried out for determining the target learners LearningObjects forindividual use are selected asthe main elements of theactivities and the criteria forselection are again a match ofthe objectives stated inmetadata (as stated incategory 9 of Learning Object MetadatamdashClassification) and thecompetencies required
(B) Sociocultural The expertise of the special interest group members as stated in the concepts addressed by their publicationsprojects or other activities
Informal group activities are the main approach applicablesince there is not a predefinedgroup structure but a cultureof sharing and socialisation The results of the sharing of ideas can be expressed in terms of oc-knowsAboutpredicates and social ties will be modelled through positive valence FOAF-knows(httpxmlnscomfoaf01)predicates with an added numerical value representingstrength
348 British Journal of Educational Technology Vol 37 No 3 2006
copy 2006 The Author Journal compilation copy 2006 British Educational Communications and Technology Agency
a complex kind including their intended participants educational objectives and exe-cution constraints The representation and recording of the assumptions and guidelinesthat were considered in the creation of such resources would extend current techniquesto capture the rationale of the design itself This in turn will eventually result in theavailability of a large base of empirical data available in which theoretical assumptionsor rules are explicitly connected to the resulting designs as these could be linked withthe actual results and trace of the learning processes that used them The resulting bodyof evidence would represent a critical asset in the development of the theory of learningmediated through computers
This paper has described the first attempt to extend existing learning technology modelswith a schema explicitly targeted at describing learning design rationales The OpenCycknowledge base has been used as the semantic reference from which the ontologicalschema has been specified Concretely an activity-based schema based on IMS LD hasbeen described and theoretical assumptions and codified guidelines have been intro-duced The flexibility of the representation schema has been analysed through therepresentation of two stereotyped theoretical positions
Two main research directions should continue the work described here On the onehand further ontological engineering effort is required to come up with a more com-prehensive and rich schema for learning design rationales On the other hand tool
Table 6 Main elements of the two learning design scenarios
Scenario Approach Examples
(A) Systemic-Behaviourist Formal logical sentencesare used to guarantee that every design followsan organisationalobjective
It is required that the objective of every Method is attached to a Competency-Spec instance as objective
Formal assessment is a requirement foraccountability purposes
Guideline (GD1) is attached to everyDesigningLearningResource instance that produces an Act
(B) Sociocultural The interaction in informal settings can beused to derive social models (Sicilia amp Garciacutea2004)
The interactions in discussion canbe used to measure the degree of lsquoproximityrsquo between members Therationale of the activities can bejustified in terms of SocialTies inside SocialNetworks Further the focus of each groupactivity should be clearly stated anda combination of the increase in collective knowledge and the strengthening of social relations can be used for assessment
Semantic learning designs 349
copy 2006 The Author Journal compilation copy 2006 British Educational Communications and Technology Agency
implementation and record gathering efforts would be required to attain the ultimateobjective of the ontological definitions described so far With such effort an unprece-dented research material could be available in some years providing improved oppor-tunities for scientific inquiry on design as an activity regarding pedagogy
Other potential directions for research may be the description of interpretations of thedesign intentions (for which no rationale was explicitly recorded) the mining or detec-tion of patterns as related to rationale description or the detailed use of theoreticalrationales in agents that support the learning process which represent a category ofrelevant learning support software (Holmes in press)
ReferencesAllert H (2004) Coherent social systems for learningmdashan approach for contextualized and
community-centred metadata Journal of Interactive Media in Education 2004 1Baader F Calvanese D McGuinness D Nardi D amp Patel-Schneider P (Eds) (2003) The
description logic handbook Theory implementation and applications Cambridge CambridgeUniversity Press
Bannister D (1981) Personal construct theory and research method In P Reason amp J Rowan(Eds) Human inquiry a sourcebook of new paradigm research (pp 191ndash200) Chichester UKJohn Wiley and Sons
Berners-Lee T Hendler J amp Lassila O (2001) The semantic web Scientific American 284 534ndash43
Brown A amp Voltz B (2005) Elements of effective e-learning design International Review ofResearch in Open and Distance Learning 6 1
Buzza D Bean D Harrigan K amp Carey T (2005) Learning design repositories adaptinglearning design specifications for shared instructional knowledge Canadian Journal of Learningand Technology 30 3
Conole G Dyke M Oliver M amp Seale J (2004) Mapping pedagogy and tools for effectivelearning design Computers amp Education 43 1ndash2 17ndash33
Dodero J M Aedo I amp Diacuteaz P (2002) Participative knowledge production of learning objectsfor e-books The Electronic Library 20 4 296ndash305
Friesen N (2005) Interoperability and learning objects an overview of e-learning standardiza-tion Interdisciplinary Journal of Knowledge and Learning Objects 1 1 23ndash31
Gruber T (1995) Towards principles for the design of ontologies used for knowledge sharingInternational Journal of HumanndashComputer Studies 43 56 907ndash928
Hatchuel A (2002) Towards design theory and expandable rationality the unfinished programof Herbert Simon Journal of Management and Governance 5 3ndash4 260ndash271
Holmes J (in press) Designing agents to support learning by explaining Computers and EducationIEEE Learning Technology Standards Committee (2002) Learning object metadata (LOM) Final
Draft Standard IEEE 1484121ndash2002IMS (2003) IMS Learning Design Information Model (Version 10) Final Specification 20
January 2003Kaptelinin V Kuutti K amp Bannon L (1995) Activity theory basic concepts and applications
In Blumenthalet al (Eds) Humanndashcomputer interaction Lecture Notes in Computer Science 1015189ndash201
Koper R (2004) Use of the semantic web to solve some basic problems in education increaseflexible distributed lifelong learning decrease teacherrsquos workload Journal of Interactive Mediain Education 2004 6
Lenat D (1995) CYC a large-scale investment in knowledge infrastructure Communications ofthe ACM 38 11 33ndash38
350 British Journal of Educational Technology Vol 37 No 3 2006
copy 2006 The Author Journal compilation copy 2006 British Educational Communications and Technology Agency
McGreal R (2004) Learning objects a practical definition International Journal of InstructionalTechnology and Distance Learning 1 9 21ndash32
Packer M amp Goicoechea J (2000) Sociocultural and constructivist theories of learningontology not just epistemology Educational Psychologist 35 4 227ndash241
Palmer J (2001) Fifty major thinkers on education London and New York RoutledgePolsani P R (2003) Use and abuse of reusable learning objects Journal of Digital Information 3
4 Rothwell W amp Kazanas H (1992) Mastering the instructional design process San Francisco CA
Jossey-BassSaacutenchez-Alonso S amp Sicilia M A (2004) Relationships and commitments in learning object
metadata In Proceedings of the 5th International Conference on Information Technology BasedHigher Education and Training ITHET 2004 Istanbul Turkey
Sicilia M A (2005) Ontology-Based Competency Management Infrastructures for theKnowledge-intensive Learning Organization In Lytras amp Naeve (Eds) Intelligent Learning Infra-structures in Knowledge Intensive Organizations A Semantic Web perspective IDEA USA pp 302ndash324
Sicilia M A amp Garciacutea E (2004) Fuzzy group models for adaptation in cooperative informationretrieval contexts Lecture Notes in Computer Science 2932 324ndash334
Sicilia M A amp Garciacutea E (2005) On the convergence of formal ontologies and standardizede-learning Journal of Distance Education Technologies 3 2 13ndash29
Sicilia M A amp Lytras M (2005) On the representation of change according to different ontol-ogies of learning International Journal of Learning and Change 1 1 66ndash79
Sicilia M A Lytras M Rodriacuteguez E amp Garciacutea E (in press) Integrating descriptions of knowl-edge management learning activities into large ontological structures a case study Data andKnowledge Engineering
Tennant M (1991) Expertise as a dimension of adult development New Education 13 1 49ndash55Vanderdonckt J (1999) Development milestones towards a tool for working with guidelines
Interacting with Computers 12 2 81ndash118Wand Y amp Weber R (1995) On the deep structure of information systems Information Systems
Journal 5 203ndash223Wiley D A (2001) The instructional use of learning objects Bloomington IN Association for
Educational Communications and Technology
Semantic learning designs 347
copy 2006 The Author Journal compilation copy 2006 British Educational Communications and Technology Agency
of (B) In addition case (B) requires a model of social network that is largely irrelevantin case (A)
Differences in recording activity rationalesThe approach for the recording of design rationales also differs because of the differentcontexts Table 6 summarises the main elements for each of the scenarios
Conclusions and directions for researchCurrent technology and practice related to the description of reusable digital learningresources has resulted in techniques that allow the description of learning activities of
Table 5 Inputs activities and objectives of two learning design scenarios
Scenario Design inputs Activities and their objectives
(A) Systemic-Behaviourist A formal competency recordfor employees and competencyrequirements expressed in terms of amounts of required competencies (assuming a shared measurement scale) Such requirements are actually derived from strategy or from the contingent requirements of upcoming projects
Competency instances are used to measure the levelof Competency-Spec for the required competencies (or their constituents) A formal match considering the agenda of employees is carried out for determining the target learners LearningObjects forindividual use are selected asthe main elements of theactivities and the criteria forselection are again a match ofthe objectives stated inmetadata (as stated incategory 9 of Learning Object MetadatamdashClassification) and thecompetencies required
(B) Sociocultural The expertise of the special interest group members as stated in the concepts addressed by their publicationsprojects or other activities
Informal group activities are the main approach applicablesince there is not a predefinedgroup structure but a cultureof sharing and socialisation The results of the sharing of ideas can be expressed in terms of oc-knowsAboutpredicates and social ties will be modelled through positive valence FOAF-knows(httpxmlnscomfoaf01)predicates with an added numerical value representingstrength
348 British Journal of Educational Technology Vol 37 No 3 2006
copy 2006 The Author Journal compilation copy 2006 British Educational Communications and Technology Agency
a complex kind including their intended participants educational objectives and exe-cution constraints The representation and recording of the assumptions and guidelinesthat were considered in the creation of such resources would extend current techniquesto capture the rationale of the design itself This in turn will eventually result in theavailability of a large base of empirical data available in which theoretical assumptionsor rules are explicitly connected to the resulting designs as these could be linked withthe actual results and trace of the learning processes that used them The resulting bodyof evidence would represent a critical asset in the development of the theory of learningmediated through computers
This paper has described the first attempt to extend existing learning technology modelswith a schema explicitly targeted at describing learning design rationales The OpenCycknowledge base has been used as the semantic reference from which the ontologicalschema has been specified Concretely an activity-based schema based on IMS LD hasbeen described and theoretical assumptions and codified guidelines have been intro-duced The flexibility of the representation schema has been analysed through therepresentation of two stereotyped theoretical positions
Two main research directions should continue the work described here On the onehand further ontological engineering effort is required to come up with a more com-prehensive and rich schema for learning design rationales On the other hand tool
Table 6 Main elements of the two learning design scenarios
Scenario Approach Examples
(A) Systemic-Behaviourist Formal logical sentencesare used to guarantee that every design followsan organisationalobjective
It is required that the objective of every Method is attached to a Competency-Spec instance as objective
Formal assessment is a requirement foraccountability purposes
Guideline (GD1) is attached to everyDesigningLearningResource instance that produces an Act
(B) Sociocultural The interaction in informal settings can beused to derive social models (Sicilia amp Garciacutea2004)
The interactions in discussion canbe used to measure the degree of lsquoproximityrsquo between members Therationale of the activities can bejustified in terms of SocialTies inside SocialNetworks Further the focus of each groupactivity should be clearly stated anda combination of the increase in collective knowledge and the strengthening of social relations can be used for assessment
Semantic learning designs 349
copy 2006 The Author Journal compilation copy 2006 British Educational Communications and Technology Agency
implementation and record gathering efforts would be required to attain the ultimateobjective of the ontological definitions described so far With such effort an unprece-dented research material could be available in some years providing improved oppor-tunities for scientific inquiry on design as an activity regarding pedagogy
Other potential directions for research may be the description of interpretations of thedesign intentions (for which no rationale was explicitly recorded) the mining or detec-tion of patterns as related to rationale description or the detailed use of theoreticalrationales in agents that support the learning process which represent a category ofrelevant learning support software (Holmes in press)
ReferencesAllert H (2004) Coherent social systems for learningmdashan approach for contextualized and
community-centred metadata Journal of Interactive Media in Education 2004 1Baader F Calvanese D McGuinness D Nardi D amp Patel-Schneider P (Eds) (2003) The
description logic handbook Theory implementation and applications Cambridge CambridgeUniversity Press
Bannister D (1981) Personal construct theory and research method In P Reason amp J Rowan(Eds) Human inquiry a sourcebook of new paradigm research (pp 191ndash200) Chichester UKJohn Wiley and Sons
Berners-Lee T Hendler J amp Lassila O (2001) The semantic web Scientific American 284 534ndash43
Brown A amp Voltz B (2005) Elements of effective e-learning design International Review ofResearch in Open and Distance Learning 6 1
Buzza D Bean D Harrigan K amp Carey T (2005) Learning design repositories adaptinglearning design specifications for shared instructional knowledge Canadian Journal of Learningand Technology 30 3
Conole G Dyke M Oliver M amp Seale J (2004) Mapping pedagogy and tools for effectivelearning design Computers amp Education 43 1ndash2 17ndash33
Dodero J M Aedo I amp Diacuteaz P (2002) Participative knowledge production of learning objectsfor e-books The Electronic Library 20 4 296ndash305
Friesen N (2005) Interoperability and learning objects an overview of e-learning standardiza-tion Interdisciplinary Journal of Knowledge and Learning Objects 1 1 23ndash31
Gruber T (1995) Towards principles for the design of ontologies used for knowledge sharingInternational Journal of HumanndashComputer Studies 43 56 907ndash928
Hatchuel A (2002) Towards design theory and expandable rationality the unfinished programof Herbert Simon Journal of Management and Governance 5 3ndash4 260ndash271
Holmes J (in press) Designing agents to support learning by explaining Computers and EducationIEEE Learning Technology Standards Committee (2002) Learning object metadata (LOM) Final
Draft Standard IEEE 1484121ndash2002IMS (2003) IMS Learning Design Information Model (Version 10) Final Specification 20
January 2003Kaptelinin V Kuutti K amp Bannon L (1995) Activity theory basic concepts and applications
In Blumenthalet al (Eds) Humanndashcomputer interaction Lecture Notes in Computer Science 1015189ndash201
Koper R (2004) Use of the semantic web to solve some basic problems in education increaseflexible distributed lifelong learning decrease teacherrsquos workload Journal of Interactive Mediain Education 2004 6
Lenat D (1995) CYC a large-scale investment in knowledge infrastructure Communications ofthe ACM 38 11 33ndash38
350 British Journal of Educational Technology Vol 37 No 3 2006
copy 2006 The Author Journal compilation copy 2006 British Educational Communications and Technology Agency
McGreal R (2004) Learning objects a practical definition International Journal of InstructionalTechnology and Distance Learning 1 9 21ndash32
Packer M amp Goicoechea J (2000) Sociocultural and constructivist theories of learningontology not just epistemology Educational Psychologist 35 4 227ndash241
Palmer J (2001) Fifty major thinkers on education London and New York RoutledgePolsani P R (2003) Use and abuse of reusable learning objects Journal of Digital Information 3
4 Rothwell W amp Kazanas H (1992) Mastering the instructional design process San Francisco CA
Jossey-BassSaacutenchez-Alonso S amp Sicilia M A (2004) Relationships and commitments in learning object
metadata In Proceedings of the 5th International Conference on Information Technology BasedHigher Education and Training ITHET 2004 Istanbul Turkey
Sicilia M A (2005) Ontology-Based Competency Management Infrastructures for theKnowledge-intensive Learning Organization In Lytras amp Naeve (Eds) Intelligent Learning Infra-structures in Knowledge Intensive Organizations A Semantic Web perspective IDEA USA pp 302ndash324
Sicilia M A amp Garciacutea E (2004) Fuzzy group models for adaptation in cooperative informationretrieval contexts Lecture Notes in Computer Science 2932 324ndash334
Sicilia M A amp Garciacutea E (2005) On the convergence of formal ontologies and standardizede-learning Journal of Distance Education Technologies 3 2 13ndash29
Sicilia M A amp Lytras M (2005) On the representation of change according to different ontol-ogies of learning International Journal of Learning and Change 1 1 66ndash79
Sicilia M A Lytras M Rodriacuteguez E amp Garciacutea E (in press) Integrating descriptions of knowl-edge management learning activities into large ontological structures a case study Data andKnowledge Engineering
Tennant M (1991) Expertise as a dimension of adult development New Education 13 1 49ndash55Vanderdonckt J (1999) Development milestones towards a tool for working with guidelines
Interacting with Computers 12 2 81ndash118Wand Y amp Weber R (1995) On the deep structure of information systems Information Systems
Journal 5 203ndash223Wiley D A (2001) The instructional use of learning objects Bloomington IN Association for
Educational Communications and Technology
348 British Journal of Educational Technology Vol 37 No 3 2006
copy 2006 The Author Journal compilation copy 2006 British Educational Communications and Technology Agency
a complex kind including their intended participants educational objectives and exe-cution constraints The representation and recording of the assumptions and guidelinesthat were considered in the creation of such resources would extend current techniquesto capture the rationale of the design itself This in turn will eventually result in theavailability of a large base of empirical data available in which theoretical assumptionsor rules are explicitly connected to the resulting designs as these could be linked withthe actual results and trace of the learning processes that used them The resulting bodyof evidence would represent a critical asset in the development of the theory of learningmediated through computers
This paper has described the first attempt to extend existing learning technology modelswith a schema explicitly targeted at describing learning design rationales The OpenCycknowledge base has been used as the semantic reference from which the ontologicalschema has been specified Concretely an activity-based schema based on IMS LD hasbeen described and theoretical assumptions and codified guidelines have been intro-duced The flexibility of the representation schema has been analysed through therepresentation of two stereotyped theoretical positions
Two main research directions should continue the work described here On the onehand further ontological engineering effort is required to come up with a more com-prehensive and rich schema for learning design rationales On the other hand tool
Table 6 Main elements of the two learning design scenarios
Scenario Approach Examples
(A) Systemic-Behaviourist Formal logical sentencesare used to guarantee that every design followsan organisationalobjective
It is required that the objective of every Method is attached to a Competency-Spec instance as objective
Formal assessment is a requirement foraccountability purposes
Guideline (GD1) is attached to everyDesigningLearningResource instance that produces an Act
(B) Sociocultural The interaction in informal settings can beused to derive social models (Sicilia amp Garciacutea2004)
The interactions in discussion canbe used to measure the degree of lsquoproximityrsquo between members Therationale of the activities can bejustified in terms of SocialTies inside SocialNetworks Further the focus of each groupactivity should be clearly stated anda combination of the increase in collective knowledge and the strengthening of social relations can be used for assessment
Semantic learning designs 349
copy 2006 The Author Journal compilation copy 2006 British Educational Communications and Technology Agency
implementation and record gathering efforts would be required to attain the ultimateobjective of the ontological definitions described so far With such effort an unprece-dented research material could be available in some years providing improved oppor-tunities for scientific inquiry on design as an activity regarding pedagogy
Other potential directions for research may be the description of interpretations of thedesign intentions (for which no rationale was explicitly recorded) the mining or detec-tion of patterns as related to rationale description or the detailed use of theoreticalrationales in agents that support the learning process which represent a category ofrelevant learning support software (Holmes in press)
ReferencesAllert H (2004) Coherent social systems for learningmdashan approach for contextualized and
community-centred metadata Journal of Interactive Media in Education 2004 1Baader F Calvanese D McGuinness D Nardi D amp Patel-Schneider P (Eds) (2003) The
description logic handbook Theory implementation and applications Cambridge CambridgeUniversity Press
Bannister D (1981) Personal construct theory and research method In P Reason amp J Rowan(Eds) Human inquiry a sourcebook of new paradigm research (pp 191ndash200) Chichester UKJohn Wiley and Sons
Berners-Lee T Hendler J amp Lassila O (2001) The semantic web Scientific American 284 534ndash43
Brown A amp Voltz B (2005) Elements of effective e-learning design International Review ofResearch in Open and Distance Learning 6 1
Buzza D Bean D Harrigan K amp Carey T (2005) Learning design repositories adaptinglearning design specifications for shared instructional knowledge Canadian Journal of Learningand Technology 30 3
Conole G Dyke M Oliver M amp Seale J (2004) Mapping pedagogy and tools for effectivelearning design Computers amp Education 43 1ndash2 17ndash33
Dodero J M Aedo I amp Diacuteaz P (2002) Participative knowledge production of learning objectsfor e-books The Electronic Library 20 4 296ndash305
Friesen N (2005) Interoperability and learning objects an overview of e-learning standardiza-tion Interdisciplinary Journal of Knowledge and Learning Objects 1 1 23ndash31
Gruber T (1995) Towards principles for the design of ontologies used for knowledge sharingInternational Journal of HumanndashComputer Studies 43 56 907ndash928
Hatchuel A (2002) Towards design theory and expandable rationality the unfinished programof Herbert Simon Journal of Management and Governance 5 3ndash4 260ndash271
Holmes J (in press) Designing agents to support learning by explaining Computers and EducationIEEE Learning Technology Standards Committee (2002) Learning object metadata (LOM) Final
Draft Standard IEEE 1484121ndash2002IMS (2003) IMS Learning Design Information Model (Version 10) Final Specification 20
January 2003Kaptelinin V Kuutti K amp Bannon L (1995) Activity theory basic concepts and applications
In Blumenthalet al (Eds) Humanndashcomputer interaction Lecture Notes in Computer Science 1015189ndash201
Koper R (2004) Use of the semantic web to solve some basic problems in education increaseflexible distributed lifelong learning decrease teacherrsquos workload Journal of Interactive Mediain Education 2004 6
Lenat D (1995) CYC a large-scale investment in knowledge infrastructure Communications ofthe ACM 38 11 33ndash38
350 British Journal of Educational Technology Vol 37 No 3 2006
copy 2006 The Author Journal compilation copy 2006 British Educational Communications and Technology Agency
McGreal R (2004) Learning objects a practical definition International Journal of InstructionalTechnology and Distance Learning 1 9 21ndash32
Packer M amp Goicoechea J (2000) Sociocultural and constructivist theories of learningontology not just epistemology Educational Psychologist 35 4 227ndash241
Palmer J (2001) Fifty major thinkers on education London and New York RoutledgePolsani P R (2003) Use and abuse of reusable learning objects Journal of Digital Information 3
4 Rothwell W amp Kazanas H (1992) Mastering the instructional design process San Francisco CA
Jossey-BassSaacutenchez-Alonso S amp Sicilia M A (2004) Relationships and commitments in learning object
metadata In Proceedings of the 5th International Conference on Information Technology BasedHigher Education and Training ITHET 2004 Istanbul Turkey
Sicilia M A (2005) Ontology-Based Competency Management Infrastructures for theKnowledge-intensive Learning Organization In Lytras amp Naeve (Eds) Intelligent Learning Infra-structures in Knowledge Intensive Organizations A Semantic Web perspective IDEA USA pp 302ndash324
Sicilia M A amp Garciacutea E (2004) Fuzzy group models for adaptation in cooperative informationretrieval contexts Lecture Notes in Computer Science 2932 324ndash334
Sicilia M A amp Garciacutea E (2005) On the convergence of formal ontologies and standardizede-learning Journal of Distance Education Technologies 3 2 13ndash29
Sicilia M A amp Lytras M (2005) On the representation of change according to different ontol-ogies of learning International Journal of Learning and Change 1 1 66ndash79
Sicilia M A Lytras M Rodriacuteguez E amp Garciacutea E (in press) Integrating descriptions of knowl-edge management learning activities into large ontological structures a case study Data andKnowledge Engineering
Tennant M (1991) Expertise as a dimension of adult development New Education 13 1 49ndash55Vanderdonckt J (1999) Development milestones towards a tool for working with guidelines
Interacting with Computers 12 2 81ndash118Wand Y amp Weber R (1995) On the deep structure of information systems Information Systems
Journal 5 203ndash223Wiley D A (2001) The instructional use of learning objects Bloomington IN Association for
Educational Communications and Technology
Semantic learning designs 349
copy 2006 The Author Journal compilation copy 2006 British Educational Communications and Technology Agency
implementation and record gathering efforts would be required to attain the ultimateobjective of the ontological definitions described so far With such effort an unprece-dented research material could be available in some years providing improved oppor-tunities for scientific inquiry on design as an activity regarding pedagogy
Other potential directions for research may be the description of interpretations of thedesign intentions (for which no rationale was explicitly recorded) the mining or detec-tion of patterns as related to rationale description or the detailed use of theoreticalrationales in agents that support the learning process which represent a category ofrelevant learning support software (Holmes in press)
ReferencesAllert H (2004) Coherent social systems for learningmdashan approach for contextualized and
community-centred metadata Journal of Interactive Media in Education 2004 1Baader F Calvanese D McGuinness D Nardi D amp Patel-Schneider P (Eds) (2003) The
description logic handbook Theory implementation and applications Cambridge CambridgeUniversity Press
Bannister D (1981) Personal construct theory and research method In P Reason amp J Rowan(Eds) Human inquiry a sourcebook of new paradigm research (pp 191ndash200) Chichester UKJohn Wiley and Sons
Berners-Lee T Hendler J amp Lassila O (2001) The semantic web Scientific American 284 534ndash43
Brown A amp Voltz B (2005) Elements of effective e-learning design International Review ofResearch in Open and Distance Learning 6 1
Buzza D Bean D Harrigan K amp Carey T (2005) Learning design repositories adaptinglearning design specifications for shared instructional knowledge Canadian Journal of Learningand Technology 30 3
Conole G Dyke M Oliver M amp Seale J (2004) Mapping pedagogy and tools for effectivelearning design Computers amp Education 43 1ndash2 17ndash33
Dodero J M Aedo I amp Diacuteaz P (2002) Participative knowledge production of learning objectsfor e-books The Electronic Library 20 4 296ndash305
Friesen N (2005) Interoperability and learning objects an overview of e-learning standardiza-tion Interdisciplinary Journal of Knowledge and Learning Objects 1 1 23ndash31
Gruber T (1995) Towards principles for the design of ontologies used for knowledge sharingInternational Journal of HumanndashComputer Studies 43 56 907ndash928
Hatchuel A (2002) Towards design theory and expandable rationality the unfinished programof Herbert Simon Journal of Management and Governance 5 3ndash4 260ndash271
Holmes J (in press) Designing agents to support learning by explaining Computers and EducationIEEE Learning Technology Standards Committee (2002) Learning object metadata (LOM) Final
Draft Standard IEEE 1484121ndash2002IMS (2003) IMS Learning Design Information Model (Version 10) Final Specification 20
January 2003Kaptelinin V Kuutti K amp Bannon L (1995) Activity theory basic concepts and applications
In Blumenthalet al (Eds) Humanndashcomputer interaction Lecture Notes in Computer Science 1015189ndash201
Koper R (2004) Use of the semantic web to solve some basic problems in education increaseflexible distributed lifelong learning decrease teacherrsquos workload Journal of Interactive Mediain Education 2004 6
Lenat D (1995) CYC a large-scale investment in knowledge infrastructure Communications ofthe ACM 38 11 33ndash38
350 British Journal of Educational Technology Vol 37 No 3 2006
copy 2006 The Author Journal compilation copy 2006 British Educational Communications and Technology Agency
McGreal R (2004) Learning objects a practical definition International Journal of InstructionalTechnology and Distance Learning 1 9 21ndash32
Packer M amp Goicoechea J (2000) Sociocultural and constructivist theories of learningontology not just epistemology Educational Psychologist 35 4 227ndash241
Palmer J (2001) Fifty major thinkers on education London and New York RoutledgePolsani P R (2003) Use and abuse of reusable learning objects Journal of Digital Information 3
4 Rothwell W amp Kazanas H (1992) Mastering the instructional design process San Francisco CA
Jossey-BassSaacutenchez-Alonso S amp Sicilia M A (2004) Relationships and commitments in learning object
metadata In Proceedings of the 5th International Conference on Information Technology BasedHigher Education and Training ITHET 2004 Istanbul Turkey
Sicilia M A (2005) Ontology-Based Competency Management Infrastructures for theKnowledge-intensive Learning Organization In Lytras amp Naeve (Eds) Intelligent Learning Infra-structures in Knowledge Intensive Organizations A Semantic Web perspective IDEA USA pp 302ndash324
Sicilia M A amp Garciacutea E (2004) Fuzzy group models for adaptation in cooperative informationretrieval contexts Lecture Notes in Computer Science 2932 324ndash334
Sicilia M A amp Garciacutea E (2005) On the convergence of formal ontologies and standardizede-learning Journal of Distance Education Technologies 3 2 13ndash29
Sicilia M A amp Lytras M (2005) On the representation of change according to different ontol-ogies of learning International Journal of Learning and Change 1 1 66ndash79
Sicilia M A Lytras M Rodriacuteguez E amp Garciacutea E (in press) Integrating descriptions of knowl-edge management learning activities into large ontological structures a case study Data andKnowledge Engineering
Tennant M (1991) Expertise as a dimension of adult development New Education 13 1 49ndash55Vanderdonckt J (1999) Development milestones towards a tool for working with guidelines
Interacting with Computers 12 2 81ndash118Wand Y amp Weber R (1995) On the deep structure of information systems Information Systems
Journal 5 203ndash223Wiley D A (2001) The instructional use of learning objects Bloomington IN Association for
Educational Communications and Technology
350 British Journal of Educational Technology Vol 37 No 3 2006
copy 2006 The Author Journal compilation copy 2006 British Educational Communications and Technology Agency
McGreal R (2004) Learning objects a practical definition International Journal of InstructionalTechnology and Distance Learning 1 9 21ndash32
Packer M amp Goicoechea J (2000) Sociocultural and constructivist theories of learningontology not just epistemology Educational Psychologist 35 4 227ndash241
Palmer J (2001) Fifty major thinkers on education London and New York RoutledgePolsani P R (2003) Use and abuse of reusable learning objects Journal of Digital Information 3
4 Rothwell W amp Kazanas H (1992) Mastering the instructional design process San Francisco CA
Jossey-BassSaacutenchez-Alonso S amp Sicilia M A (2004) Relationships and commitments in learning object
metadata In Proceedings of the 5th International Conference on Information Technology BasedHigher Education and Training ITHET 2004 Istanbul Turkey
Sicilia M A (2005) Ontology-Based Competency Management Infrastructures for theKnowledge-intensive Learning Organization In Lytras amp Naeve (Eds) Intelligent Learning Infra-structures in Knowledge Intensive Organizations A Semantic Web perspective IDEA USA pp 302ndash324
Sicilia M A amp Garciacutea E (2004) Fuzzy group models for adaptation in cooperative informationretrieval contexts Lecture Notes in Computer Science 2932 324ndash334
Sicilia M A amp Garciacutea E (2005) On the convergence of formal ontologies and standardizede-learning Journal of Distance Education Technologies 3 2 13ndash29
Sicilia M A amp Lytras M (2005) On the representation of change according to different ontol-ogies of learning International Journal of Learning and Change 1 1 66ndash79
Sicilia M A Lytras M Rodriacuteguez E amp Garciacutea E (in press) Integrating descriptions of knowl-edge management learning activities into large ontological structures a case study Data andKnowledge Engineering
Tennant M (1991) Expertise as a dimension of adult development New Education 13 1 49ndash55Vanderdonckt J (1999) Development milestones towards a tool for working with guidelines
Interacting with Computers 12 2 81ndash118Wand Y amp Weber R (1995) On the deep structure of information systems Information Systems
Journal 5 203ndash223Wiley D A (2001) The instructional use of learning objects Bloomington IN Association for
Educational Communications and Technology