1 Knowledge Exchange Franz J. Kurfess Cal Poly SLO Computer Science Department.

55
1 Knowledge Exchange Franz J. Kurfess Cal Poly SLO Computer Science Department
  • date post

    20-Dec-2015
  • Category

    Documents

  • view

    217
  • download

    0

Transcript of 1 Knowledge Exchange Franz J. Kurfess Cal Poly SLO Computer Science Department.

Page 1: 1 Knowledge Exchange Franz J. Kurfess Cal Poly SLO Computer Science Department.

1

Knowledge Exchange

Franz J. Kurfess

Cal Poly SLOComputer Science Department

Page 2: 1 Knowledge Exchange Franz J. Kurfess Cal Poly SLO Computer Science Department.

3

Acknowledgements

Some of the material in these slides was developed for a lecture series sponsored by the

European Community under the BPD programwith Vilnius University

as host institution

Page 3: 1 Knowledge Exchange Franz J. Kurfess Cal Poly SLO Computer Science Department.

4© Franz J. Kurfess

Use and Distribution of these Slides

❖These slides are primarily intended for the students in classes I teach. In some cases, I only make PDF versions publicly available. If you would like to get a copy of the originals (Apple KeyNote or Microsoft PowerPoint), please contact me via email at [email protected]. I hereby grant permission to use them in educational settings. If you do so, it would be nice to send me an email about it. If you’re considering using them in a commercial environment, please contact me first.

Page 4: 1 Knowledge Exchange Franz J. Kurfess Cal Poly SLO Computer Science Department.

5© Franz J. Kurfess

Overview Knowledge Exchange

❖ Introduction

❖Knowledge Capture Explicit Capture Extraction From Text Case-based Reasoning Enhancement of Existing Documents

❖Transfer of Knowledge Communication

Basic ConceptsLanguage and Communication

Natural Language Formal Languages Communication Models

❖Distribution of Knowledge Knowledge Repositories

Distribution Models

Page 5: 1 Knowledge Exchange Franz J. Kurfess Cal Poly SLO Computer Science Department.

13

Introduction

Page 6: 1 Knowledge Exchange Franz J. Kurfess Cal Poly SLO Computer Science Department.

14

© Franz J. Kurfess

((( )) ())))

Richer representationsMore ambiguousMore versatile

(defconcept bridge ()))

More formalMore concreteMore introspectible

Know

ledge Base

Introductory texts, expert hints, explanations, dialogues, comments, examples, exceptions,...

Info. extraction templates,dialogue segments and pegs,filled-out forms, high-level connections,...

Alternative formalizations (KIF, MELD, CML,…), alternative views of the same notion (e.g., what is a threat)

Descriptions augmented with prototypical examples & exceptions, problem-solving steps and substeps, ...

WWW

[Gil 2000]

The Need for Knowledge Exchange

Page 7: 1 Knowledge Exchange Franz J. Kurfess Cal Poly SLO Computer Science Department.

15

© Franz J. Kurfess[Gil 2000]

Knowledge Mobility

❖ multiple views and versions of the same information need to provide tools that establish connections among

alternative versions/views of the same information

❖ hyper-connectivity need to provide tools that suggest further connections to

related sources when users compose documents need to annotate hyperlinks

❖ basis to support information morphing how one or more knowledge sources are used for

alternative purposes track alternative knowledge transformations various renderings and implementations of a knowledge source

Page 8: 1 Knowledge Exchange Franz J. Kurfess Cal Poly SLO Computer Science Department.

16

© Franz J. Kurfess

Knowledge Capture

❖before knowledge can be shared, it must be captured in and encoding and representation that is suitable for

the sender and recipient of the knowledge the representation format must be suitable for

transmission via a communication channel between sender and recipient

for human-to-human knowledge exchange, natural language in written or spoken form is often suitable and convenient

Page 9: 1 Knowledge Exchange Franz J. Kurfess Cal Poly SLO Computer Science Department.

17

© Franz J. Kurfess

Knowledge Capture Methods

❖Explicit Capture

❖Extraction from Documents text other formats

❖Case-based Reasoning

❖Enhancement of Existing Documents

Page 10: 1 Knowledge Exchange Franz J. Kurfess Cal Poly SLO Computer Science Department.

18

© Franz J. Kurfess

Explicit Capture

❖conventional techniques for knowledge acquisition interviews with experts, knowledge engineers

❖advantages carefully constructed suitable knowledge representation methods usually common-sense evaluation sometimes formal evaluation

consistency checks, other formal aspects

Page 11: 1 Knowledge Exchange Franz J. Kurfess Cal Poly SLO Computer Science Department.

19

© Franz J. Kurfess

Extraction From Text

❖syntactic level keywords, descriptive features

construction of an index, meta data

❖semantic level document structure

requires information about structure (tags, DDT, RDF) sentence structure

natural language processing (NLP)

❖pragmatic level context (thesaurus, ontology, NLP)

Page 12: 1 Knowledge Exchange Franz J. Kurfess Cal Poly SLO Computer Science Department.

20

© Franz J. Kurfess

Case-based Reasoning

❖solutions to a problem in a specific context are collected

❖ represented in a structured format problem, context, solution usable by a computer-based system cases are often represented through frames or similar

mechanisms

❖new cases are matched against existing ones patterns in the frames provide the basis for matching the suitability of the solution is judged by the user

Page 13: 1 Knowledge Exchange Franz J. Kurfess Cal Poly SLO Computer Science Department.

21

© Franz J. Kurfess

Enhancement of Existing Documents

❖ in addition to the methods mentioned above, collections of documents can be enhanced addition of meta-knowledge integration into an existing framework/ontology

manually through categorizationautomatically through keyword extraction indirectly through statistical correlations with other

documents

Page 14: 1 Knowledge Exchange Franz J. Kurfess Cal Poly SLO Computer Science Department.

22

© Franz J. Kurfess

Transfer of Knowledge

❖Communication Basic Concepts Language and Communication

Natural LanguageFormal Languages

Communication Models

Page 15: 1 Knowledge Exchange Franz J. Kurfess Cal Poly SLO Computer Science Department.

23

© Franz J. Kurfess

Basic Concepts

❖communication exchange of information requires a shared system of signs greatly enhanced by language speaker

produces signs as utterancesgeneral: not only spoken language

listener (hearer)perceives and interprets signs

Page 16: 1 Knowledge Exchange Franz J. Kurfess Cal Poly SLO Computer Science Department.

24

© Franz J. Kurfess

Purpose of Communication

❖sharing of information among agents or systems query other agents for information responses to queries requests or commands

actions to be performed for another agent offer

proposition for collaboration acknowledgement

confirmation of requests, offers sharing

of experiences, feelings

Page 17: 1 Knowledge Exchange Franz J. Kurfess Cal Poly SLO Computer Science Department.

25

© Franz J. Kurfess

Communication Problems

❖ intention what is the expected outcome (speaker’s perspective)

❖ timing when is a communication act appropriate

❖ selection which act is the right one

❖ language what sign system should be used

❖ interpretation will the intended meaning be conveyed to the listener

❖ ambiguity can the intention be expressed without the possibility of

misunderstandings

Page 18: 1 Knowledge Exchange Franz J. Kurfess Cal Poly SLO Computer Science Department.

26

© Franz J. Kurfess

Language and Communication

❖Natural Language used by humans evolves over time moderately to highly ambiguous

❖Formal Languages invented rigidly defined little ambiguity

Page 19: 1 Knowledge Exchange Franz J. Kurfess Cal Poly SLO Computer Science Department.

27

© Franz J. Kurfess

Natural Language

❖ formal description is very difficult sometimes non-systematic, inconsistent, ambiguous

❖mostly used for human communication easy on humans tough on computers

❖context is critical situation, beliefs, goals

Page 20: 1 Knowledge Exchange Franz J. Kurfess Cal Poly SLO Computer Science Department.

28

© Franz J. Kurfess

Formal Languages

❖symbols terminal symbols

finite set of basic wordsnot: alphabet, characters

non-terminal symbols intermediate structures composed of terminal or non-

terminal symbols

❖strings sequences of symbols

❖phrases sub-strings grouping important parts of a string

Page 21: 1 Knowledge Exchange Franz J. Kurfess Cal Poly SLO Computer Science Department.

29

© Franz J. Kurfess

Formal Languages Cont.

❖sentences allowable strings in a language composed from phrases

❖grammar rules describing correct sentences often captured as rewrite rules in BNF notation

❖ lexicon list of allowable vocabulary words

Page 22: 1 Knowledge Exchange Franz J. Kurfess Cal Poly SLO Computer Science Department.

30

© Franz J. Kurfess

Communication Models

❖encoded message model a definite proposition of the speaker is encoded into

signs which are transmitted to the listener the listener tries to decode the signs to retrieve the

original proposition errors are consequences of transmission problems

❖situated language model the intended meaning of a message depends on the

signals as well as the situation in which they are exchanged

mis-interpretation may lead to additional problems

Page 23: 1 Knowledge Exchange Franz J. Kurfess Cal Poly SLO Computer Science Department.

31

© Franz J. Kurfess

Communication Types

❖telepathic communication speaker and listener have a shared internal

representation communication through Tell/Ask directives

❖ language-based communication speaker performs actions that produce signs which

other agents can perceive and interpret communication language is different from the internal

representation more complex

involves several mappings language needs to be generated, encoded, transmitted, decoded,

and interpreted

Page 24: 1 Knowledge Exchange Franz J. Kurfess Cal Poly SLO Computer Science Department.

32

© Franz J. Kurfess

Telepathic Communication

[Russell & Norvig 1995]

Page 25: 1 Knowledge Exchange Franz J. Kurfess Cal Poly SLO Computer Science Department.

33

© Franz J. Kurfess

Language-Based Communication

[Russell & Norvig 1995]

Page 26: 1 Knowledge Exchange Franz J. Kurfess Cal Poly SLO Computer Science Department.

34

© Franz J. Kurfess

Communication Steps: Speaker

❖ intention decision about producing a speech act

❖generation conversion of the information to be transferred into the

chosen language

❖synthesis actions that produce the generated signs

Page 27: 1 Knowledge Exchange Franz J. Kurfess Cal Poly SLO Computer Science Department.

35

© Franz J. Kurfess

Communication Steps: Listener

❖perception reception of the signs produced by the speaker

speech recognition, lip reading, character recognition analysis

syntactic interpretation (parsing)semantic interpretation

disambiguationselection of the most probable intended meaning

incorporationthe selected interpretation is added to the existing world

model as additional piece of evidence

Page 28: 1 Knowledge Exchange Franz J. Kurfess Cal Poly SLO Computer Science Department.

36

© Franz J. Kurfess

Communication Example

[Russell & Norvig 1995]36

Page 29: 1 Knowledge Exchange Franz J. Kurfess Cal Poly SLO Computer Science Department.

37

Knowledge Exchange

Perspectives

Page 30: 1 Knowledge Exchange Franz J. Kurfess Cal Poly SLO Computer Science Department.

38

© Franz J. Kurfess

Different Perspectives

❖Roles

❖Scope

❖Purpose

Page 31: 1 Knowledge Exchange Franz J. Kurfess Cal Poly SLO Computer Science Department.

39

© Franz J. Kurfess

Roles

❖ knowledge creator source of the knowledge

❖ knowledge facilitator supports the exchange of knowledge between creator and

user

❖ knowledge user

❖ sender initiates and conducts the transmission of knowledge may be creator or facilitator

❖ recipient typically the knowledge user

Page 32: 1 Knowledge Exchange Franz J. Kurfess Cal Poly SLO Computer Science Department.

40

© Franz J. Kurfess

Scope of Knowledge Exchange

❖number of people involved individuals, groups, organizations, humanity

❖coherence domain knowledge, educational background,

intellectual ability, familiarity with the environment, ...

❖spread geographical distribution

Page 33: 1 Knowledge Exchange Franz J. Kurfess Cal Poly SLO Computer Science Department.

41

© Franz J. Kurfess

Individuals

❖ informal

❖direct communication

❖quick feedback

❖ low persistence tolerable

❖clarification easy

❖consistency issues easy to resolve

Page 34: 1 Knowledge Exchange Franz J. Kurfess Cal Poly SLO Computer Science Department.

42

© Franz J. Kurfess

Groups

❖ informal

❖direct communication

❖coordination and synchronization required

❖moderate persistence desirable

❖clarification via discussion

Page 35: 1 Knowledge Exchange Franz J. Kurfess Cal Poly SLO Computer Science Department.

43

© Franz J. Kurfess

Organization

❖more formal repositories and exchange methods

❖systematic communication, coordination and synchronization necessary

❖persistence is important

❖more structured approaches to clarification and consistency beneficial

Page 36: 1 Knowledge Exchange Franz J. Kurfess Cal Poly SLO Computer Science Department.

44

© Franz J. Kurfess

Community

❖ formal “body of knowledge” well-structured, reasonably controlled vocabulary,

established repositories of knowledge, procedures for validation (“peer review”)

❖established exchange methods journals, official publications, books, conferences,

portals

❖professional organizations with controlled memberships established communication, coordination, and

synchronization methods

Page 37: 1 Knowledge Exchange Franz J. Kurfess Cal Poly SLO Computer Science Department.

45

© Franz J. Kurfess

Humanity

❖no coherent “body of knowledge”

❖communication, coordination, and synchronization of knowledge exchange across boundaries is difficult

❖differences in vocabulary, methods, knowledge validation processes make exchange of knowledge difficult

❖serious problems with clarification, resolution of inconsistencies are possible

Page 38: 1 Knowledge Exchange Franz J. Kurfess Cal Poly SLO Computer Science Department.

46

© Franz J. Kurfess

Purpose

❖personal enrichment

❖better product

❖better working conditions

❖commercial advantage

❖stronger community

❖societal benefits

Page 39: 1 Knowledge Exchange Franz J. Kurfess Cal Poly SLO Computer Science Department.

47

Knowledge Distribution

Page 40: 1 Knowledge Exchange Franz J. Kurfess Cal Poly SLO Computer Science Department.

48

© Franz J. Kurfess

Distribution of Knowledge

❖Knowledge Repositories Digital Libraries

❖Distribution Models

Page 41: 1 Knowledge Exchange Franz J. Kurfess Cal Poly SLO Computer Science Department.

49

© Franz J. Kurfess

Knowledge Repositories

❖persistent storage of digital documents internal representation in the original format

loss-less transformation may be acceptable

❖ transparent internal organization multiple presentation methods for various users and

usage methods

❖multiple access methods according to users’ needs and capabilities

Page 42: 1 Knowledge Exchange Franz J. Kurfess Cal Poly SLO Computer Science Department.

50

© Franz J. Kurfess

Wikipedia

❖collaborative effort to capture knowledge

❖contributions by volunteers not restricted to “experts”

❖ liberal policy for entry modifications editorial policies to limit abuse

Page 43: 1 Knowledge Exchange Franz J. Kurfess Cal Poly SLO Computer Science Department.

51

© Franz J. Kurfess

Scientific American “Edit This”

❖public is invited to comment on some articles before they are published

❖examples “Science 2.0: Great New Tool, or Great Risk?” “Does the U.S. Produce Too Many Scientists?”

Page 44: 1 Knowledge Exchange Franz J. Kurfess Cal Poly SLO Computer Science Department.

52

© Franz J. Kurfess

Google Knols

❖An example:

Canine Science: How Dogs Communicate

How Dogs Talk To Each Other Using Sound, Smell and Body Language

Dogs communicate with each other in a variety of ways, using sounds, smells and body language. This knol offers a guide to some of the weird and wonderful methods canines use to get their messages across.

Contents• Introduction

• Image: wikimedia.org Pet Sounds: What Dog Noises Mean

• Barking: Some Canine Statistics

• Body Talk: Physical Language

• Talking Dirty: Smells

• Further Reading:

Page 45: 1 Knowledge Exchange Franz J. Kurfess Cal Poly SLO Computer Science Department.

53

© Franz J. Kurfess

Educational Repositories

❖Open Course Ware Initiative MIT

❖Connexions Rice University and Connexions Consortium

❖Merlot California State University System is a “Sustaining

Partner”

Page 46: 1 Knowledge Exchange Franz J. Kurfess Cal Poly SLO Computer Science Department.

54

© Franz J. Kurfess

General Repositories

❖ Instructables.com

❖Slideshare.com

❖Blogs

❖Youtube and similar sites

Page 47: 1 Knowledge Exchange Franz J. Kurfess Cal Poly SLO Computer Science Department.

55

© Franz J. Kurfess

Digital Libraries

❖collections of documents and artifacts stored and accessed via computers

❖ remotely accessible through networks

❖enhanced functionality compared with paper-based libraries access methods organization principles duplication

❖ implementation and usage unclear

Page 48: 1 Knowledge Exchange Franz J. Kurfess Cal Poly SLO Computer Science Department.

56

© Franz J. Kurfess

Digital Library In-A-Box

[Sweeney & Kurfess 1998]

Page 49: 1 Knowledge Exchange Franz J. Kurfess Cal Poly SLO Computer Science Department.

57

Usability Aspects of Knowledge Exchange

Page 50: 1 Knowledge Exchange Franz J. Kurfess Cal Poly SLO Computer Science Department.

58

© Franz J. Kurfess

Model Mismatch

❖sender and recipient rely on different conceptual models mis-interpretation of words, statements, intentions additional efforts required to align conceptual models

Page 51: 1 Knowledge Exchange Franz J. Kurfess Cal Poly SLO Computer Science Department.

59

© Franz J. Kurfess

Level Mismatch

❖sender and recipient use different levels of detail recipient may be bored because there is nothing of

interest in the sender’s statements recipient may be overwhelmed by the details

Page 52: 1 Knowledge Exchange Franz J. Kurfess Cal Poly SLO Computer Science Department.

60

© Franz J. Kurfess

Intention Mismatch

❖ the intention with which a sender makes a statement is different from what the recipient assumes

Page 53: 1 Knowledge Exchange Franz J. Kurfess Cal Poly SLO Computer Science Department.

61

© Franz J. Kurfess

Presentation Mismatch

❖ the way statements are presented to a recipient is not suited recipient does not have the capabilities to perceive the

information e.g. showing a picture to someone who is blind

context is unsuited for the presentation method speech in a noisy environment visual information when the recipient’s attention is engaged

elsewhere navigation information while driving

the presentation method makes it difficult for the recipient to identify relevant aspects information overload recall vs. recognition

Page 54: 1 Knowledge Exchange Franz J. Kurfess Cal Poly SLO Computer Science Department.

62

© Franz J. Kurfess

Transmission Problems

❖ the message to be communicated does not arrive as intended damaged, lost may be related to presentation mismatch

transmission method or medium is unsuitable for the context

Page 55: 1 Knowledge Exchange Franz J. Kurfess Cal Poly SLO Computer Science Department.

70

© Franz J. Kurfess