Modeling Behaviour of the Users in Adaptive and Semantic-enhanced

Information Systems: The Role of a User Ontology

Dr. Liana Razmerita

Centre of Applied Information and Communication Technologies Copenhagen Business School

Liana Razmerita, Adaptive Hypermedia, 29/07/08

2

Presentation Outline

IntroductionContext of research - KMSMotivation- painpoints of the actual KM toolsResearch Questions

Ontology-based User Modeling Conclusions and Outlook

Liana Razmerita, Adaptive Hypermedia, 29/07/08

3

Context of researchKnowledge Management Systems (KMS)- information systems

dedicated to manage organizational knowledge (Leidner and Alavi, 2001).

Semantic-enhanced KMSs - Domain ontologies can improve KMS within organizations and among distributed web-comunities (Maedche et al., 2002, Velardi et al, 2007)

create

capitalize

share

KMS integrate complex knowledge processes:

• Support social processes: ks, kc.• Collaboration between employees;• Learning processes;• Management of tacit knowledge;

KM 1.0, KM 2.0, KM3.0 –semantic-enhanced

Liana Razmerita, Adaptive Hypermedia, 29/07/08

4

Limitations of current KMSs

Need to better organize knowledge “Content is not correctly organized, not updated or [is] duplicated.”

Need for better knowledge filtering and user support “These tools need major improvement to allow users [to]

use knowledge tools in an easy way, spend less time and avoid getting lost among hundreds of document”,

”Save time when I am looking for a solution”

Need to better manage tacit knowledge “To know what people know and to make their

experience with technology and products accessible”.

Source: Survey on KMS (n=16 Ontologging End-Users )

Liana Razmerita, Adaptive Hypermedia, 29/07/08

5

Research Questions

Why is it important to model users of a KMS? What are the relevant characteristics of the users of a KMS? What type of user’s behavior can be distinguished in a KMS? How to track and maintain the user models in a KMS?

What are the advantages/limitations of applying ontologies in user modeling? How to make us of the metadata?

How can a user model improve the interaction with a KMS? What type of intelligent /personalized services can be provided

based on the user’s characteristics?

How to build semantic-enhanced data models?What are the perspectives of its use in the context of the

Semantic Web?

Liana Razmerita, Adaptive Hypermedia, 29/07/08

6

Related WorkUser Modeling Research

Adaptive hypermedia and personalized interactions (Kobsa, 2002),(Brusilovsky, 2007),(Cristea et al., 2007)

Ontology-based user modeling (Razmerita et al., 2003)(Kay and Lum, 2005)(Heckman et al., 2005, 2007) (Katifori et al., 2008)

Semantic Web and Ontology Research Ontology-based personalization: Dolog and Nejdl(2007) EPOS

(Schwarz and Roth-Berghofer,2002), Elena(Dolog et. al., 2003); (Henze et al. 2004)

Ontology, agents and corporate memory: FRODO(Van Elst & Abecker, 2001), CoMMA(Gandon and Dieng-Kuntz, 2001)

Knowledge Management Systems (O’Leary &Studer, 2001), (Nonaka et al. 2001), (Fischer &

Ostwald, 2001)

Liana Razmerita, Adaptive Hypermedia, 29/07/08

7

From Ontology-based Knowledge Modeling to Ontology-based User ModelingIntegration of 2 different specifications for user model by 2

Spanish companiesBuilding a user ontology is not simple!

How achieve an agreed conceptualization, integrate specific characteristics of users interacting with KMS, completeness?

Adapt a methodology proposed by [Uschold and Gruninger, 1996]Step1 Specification phase

The use of IMS LIP specification Extends IMS LIP-upper ontology

Step 2 The process of coding KAON tool suite and Web ontology language extending RDF/RDFS

Step 3 The process of integrating with existing ontologies The definition of similar concepts as synonyms Different representation languages

Liana Razmerita, Adaptive Hypermedia, 29/07/08

8

OI-Modeler View of User Ontology

Behavior Concept

TypeOfActivityLevelOfActivity

LevelOfKnowledge-Sharing

Liana Razmerita, Adaptive Hypermedia, 29/07/08

9

Ontology-based User Modeling framework (OntobUMf)

User Model

User Profile

Editor

Intelligent Service I

EventsLogs

LogInstances

DomainInstances

User Ontology

Domain Ontology

Automatically

providedManually

provided

Intelligent Service n...

User Instances

Log Ontology

Documents

(Razmerita et al. 2003, Razmerita 2007)

Liana Razmerita, Adaptive Hypermedia, 29/07/08

10

Intelligent Service: category extractor

Categories of users obtained by processing the logs (heuristics + fuzzy logic);Type of Activity: Readers/ Writers/ Lurkers;

If (nb_of_read_papers>NR) and (nb_of_contributions <NC) then user(x) =”reader” (during timeframe)If (nb_of_contributions>=NC) then user(x) =”writer” (during timeframe)….

Level of Activity: Very Active/Active/Visitor/Inactive;If (nb_of_read_papers > NR) and (nb_of_contributions >=NC+1) then

user(x) = “very active“ ….If (nb_of_read_papers =0) and (nb_of_contributions =0) Then user(x)

= “inactive” (during timeframe)

Liana Razmerita, Adaptive Hypermedia, 29/07/08

11

Intelligent Service: category extractor

Level of Knowledge Sharing: Unaware, Aware, Interested, Trial, Adopters

Y=f(x1, x2)– [very high, high, medium, low, very low] x1 the type of activity: [high, medium, low]

x2 the level of activity: [high, medium, low, very low]

Y=f(x1, x2) high medium low very low

high very high very high medium

medium high medium low

low very low very low

The calculus of the level of knowledge sharing

Liana Razmerita, Adaptive Hypermedia, 29/07/08

12

Liana Razmerita, Adaptive Hypermedia, 29/07/08

13

Findings-some evaluation results

Users are concerned with privacy and trust: User’s profile should be only partial available in a

KMS; Users want to be in control and to maintain their

profiles; The use of combo box would facilitate the

acquisition of the user’s data and the consistency of the terminology;

Ontologging end-users identify recognition and promotion as key incentive for knowledge sharing

Liana Razmerita, Adaptive Hypermedia, 29/07/08

14

SummaryA top-down approach in building a user

ontology using IMS LIP specification-lack of ”know-how” on how to build an ontology?

User ontology extending IMS LIP (Information Management Systems Learning Information Package)

A standard/a specification can be usefull but can be limitative.

Ontology-based UM frameworkUser Behavior (level of activity, type of activity, level of

knowledge sharing)

Future work

Extending OntobUMf towards its use within KM2.0 (blogs, wikis)

OntobUMf in an e-learning scenario:LevelofActivity/TypeofActivity/LevelofKnowledge

Liana Razmerita, Adaptive Hypermedia, 29/07/08

15