Rinke Hoekstra<br />History of Knowledge Representation<br />10-12-2008<br />SIKS Course - Knowledge Modelling<br />1<br />
Caveat Emptor<br />About me…<br />Knowledge Engineering<br />Ontologies<br />Web Ontology Language (OWL 2)<br />Dissertati...
Overview<br />In the beginning… (400 BC – 1900s)<br />Scruffies vs. Neats (1970-ies) <br />The Dark Ages (1980-ies)<br />E...
In the beginning…<br />400BC – 1900s<br />10-12-2008<br />SIKS Course - Knowledge Modelling<br />4<br />
Aristotle (384 BC – 322 BC)<br />Dialectics <br />reductio ad absurdum<br />Deduction<br />premises  conclusion (Plato)<b...
Syllogisms<br />Example<br />Major premise	All mortal things die<br />Minor premise	All men are mortal things<br />Conclus...
Aristotle’s Categories<br />Substance<br />primary vs. secondary<br />Quantity<br />extension<br />Quality<br />nature<br ...
Porphyry of Tyre (233–c. 309)<br />10-12-2008<br />SIKS Course - Knowledge Modelling<br />8<br />
Brentano (1838-1917)<br />10-12-2008<br />SIKS Course - Knowledge Modelling<br />9<br />
Ramon Llull (1232 – 1315)<br />Mechanical aids to reasoning<br />10-12-2008<br />SIKS Course - Knowledge Modelling<br />10...
Scientific Revolution (17th and 18th century)<br />Dualism<br />René Descartes (1596 – 1650)<br />Body as machine &lt;-&gt...
John Wilkins (1614 – 1672)<br />Universal Character<br />Replace latin<br />(Metric system)<br />Tree with 3 layers<br />1...
Gottfried Wilhelm Leibniz (1646 – 1716)<br />CharacteristicaUniversalis<br />“Once the characteristic numbers of most noti...
Calculators<br />Pascaline<br />Addition<br />Substraction<br />Stepped Reckoner<br />Multiplication<br />Division<br />Bi...
Another Leibniz Quote<br />&quot;If controversies were to arise, there would be no more need of disputation between two ph...
Linnaeus (1707-1778) –SystemaNaturae<br />10-12-2008<br />SIKS Course - Knowledge Modelling<br />16<br />
… so, what’s new?<br />Syllogisms<br />Rules of valid reasoning<br />Reasoning as Calculation<br />Symbol Manipulation<br ...
GottlobFrege (1884 – 1924)<br />Logic<br />Study of correct reasoning<br />Arithmetics and Mathematics<br />Begriffschrift...
Computers<br />Algorithms<br />Alan Turing (1912 – 1954)<br />Processor/Memory Architecture<br />Neumann JánosLajos(1903 –...
Theorem Proving<br />``… great theorem proving controversy of the late sixties …’’ (Newell, 1982)<br />Problematic<br />No...
Scruffies vs. Neats<br />1970ies<br />10-12-2008<br />SIKS Course - Knowledge Modelling<br />21<br />
Two Schools (1970ies and onwards)<br />Philosophy (Neats)<br />Clean, uniform language<br />Knowledge derives from small s...
Artificial Intelligence<br />“. . . an entity is intelligent if it has an adequate modelof the world […], if it is clever ...
Epistemic and Heuristic adequacy<br />McCarthy & Hayes:<br />Representation vs. Mechanism<br />Epistemic Adequacy<br />Cor...
Heuristic vs. Epistemic views in Psychology <br />Knowledge is about the how<br />Problem Solving<br /> Production System...
Information Processing System (IPS)<br />Computer as metaphor of the mind<br />“the human operates as an <br />information...
Production Systems (1)<br />Processor<br />Interpreter<br />Elementary Information Processes (EIP)<br />Sequence of EIPs a...
Production Systems (2)<br />Adequacy?<br />Correspondence to human reasoning<br />Not ‘clean’ or ‘logical’<br />Escape lim...
Semantic Networks (1)<br />Natural Language<br />Ground lexical terms in a model of reality<br />Semantic Memory<br />M. R...
Semantic Networks (2)<br />10-12-2008<br />SIKS Course - Knowledge Modelling<br />30<br />
Semantic Networks (3)<br />Adequacy?<br />Correspondence to human memory<br />Response time<br />Property inheritance<br /...
Frames (1)<br />Criticism from Cognitive Science<br />Frames, Marvin Minsky (1975)<br />Scripts, Roger Schank (1975)<br />...
Frames (2)<br />Frame system<br />Related frames that share the same terminals<br />… different viewpoints on the same sit...
Semantic Networks (3)<br />Technical problems<br />Weak inference (inheritance)<br />Unclear semantics<br />“What’s in a l...
Frame (like) Languages<br />Emphasis<br />Interrelated, internallystructuredconcepts<br />Knowledge Representation Languag...
Knowledge Representation Language (KRL)<br />Known entity: prototype<br />Description by reusable descriptors<br />Descrip...
SI Networks<br />KL-ONE (Brachman, 1979; Brachman & Schmolze, 1985)<br />Descriptions<br />Role/Filler Descriptions<br />S...
SI-Network of an Arch<br />10-12-2008<br />SIKS Course - Knowledge Modelling<br />38<br />
Epistemological Status<br />Cognitive plausibility Epist. Status<br />Relation to reality?<br />Relation to representatio...
The Knowledge Level (Allen Newell, 1982)<br />“… the crux for AI is that no one has been able to formulate in a reasonable...
The Knowledge Level (2)<br />10-12-2008<br />SIKS Course - Knowledge Modelling<br />41<br />“There exists a distinct compu...
The Knowledge Level (3)<br />Not a stance<br />viz. the intentional stance(Dennett, 1987)<br />No representation at knowle...
Brachman’s Triangle Extended (Hoekstra, 2009)<br />10-12-2008<br />SIKS Course - Knowledge Modelling<br />43<br />
Representation and Language<br />Brachman’s levels in Semantic Nets<br />Primitives of KR languages<br />Requirements<br /...
Epistemological Level<br />Missing level<br />Knowledge-structuring primitives<br />“The formal structure of conceptual un...
Optimism<br />Modern Knowledge Representation<br />Representation of expert knowledge<br />Performance over Plausibility<b...
The Dark Ages<br />1980ies<br />10-12-2008<br />SIKS Course - Knowledge Modelling<br />47<br />
Practical Applications (1980s)<br />Expert Systems<br />Production Rules<br />Rules of thumb<br />Relatively clear status<...
MYCIN and GUIDON (William Clancey, 1983)<br />MYCIN: medical diagnosis<br />GUIDON: medical tutoring<br />“transfer back” ...
Role of Knowledge in Problem Solving<br />10-12-2008<br />SIKS Course - Knowledge Modelling<br />50<br />
Knowledge Types<br />Order of rules: problem solving strategy<br />Structure in knowledge<br />Common causes before unusua...
Convergence?<br />Heuristic vs. Epistemological Adequacy<br />Two approaches<br />Different formalisms<br />Same types of ...
Problems<br />Knowledge Acquisition Bottleneck (Feigenbaum, 1980)<br />The difficulty to correctly extract relevant knowle...
ENGINEERING REVIVAL<br />1990s<br />10-12-2008<br />SIKS Course - Knowledge Modelling<br />54<br />
Knowledge Acquisition<br />Ensure<br />Quality<br />Reuse<br />Ad hoc Methodologies<br />Engineering<br />Knowledge model...
CommonKADS(Wielinga et al., 1992, van Heijst et al., 1997)<br />Knowledge Level Model<br />Independent of KR language<br /...
Reuse<br />Role limiting<br />Direct reuse<br />Index symbol level representations<br />Detailed blueprints<br />Skeletal ...
Knowledge Types (1)<br />Control Knowledge<br />Task Knowledge<br />Inference Knowledge<br />Problem Solving Methods (Breu...
Knowledge Types (2)<br />Domain Knowledge<br />Index PSM’s for reuse  Epistemology<br />Generic domain theory<br />What a...
Functional Perspective (Hector Levesque, 1984)<br />Descend to the Symbol Level?<br />Knowledge Base<br />Abstract datatyp...
Knowledge Based Systems<br />Architecture<br />Specialised KR languages<br />Specialised Services<br />Performance guarant...
The return of logic (Levesque & Brachman, 1987)<br />Classification as logical inference<br />Exact semantics<br />Trade-o...
Description Logics (Baader & Hollunder, 1991)<br />KL-One, NIKL, KL-Two, LOOM, FL, KANDOR, KRYPTON, CLASSIC …<br />Quest<b...
… and the rest?<br />Domain Theory <br />Causal Knowledge<br />Naïve Physics<br />Qualitative Reasoning (J. de Kleer, K.D....
The ‘O’ Word<br />1995 and onwards<br />10-12-2008<br />SIKS Course - Knowledge Modelling<br />65<br />Oh no! <br />Not th...
Pop Quiz<br />What is an ontology?<br />10-12-2008<br />SIKS Course - Knowledge Modelling<br />66<br />
Ontology<br />“Ontology or the science of something and of nothing, of being and not-being of the thing and the mode of th...
Knowledge Representation (Davis, Shrobe, Szolovits, 1993)<br />Surrogate<br />Set of ontological commitments<br />through ...
Ontology Definitions<br />Knowledge Management<br />An explicit (knowledge level) specification of a conceptualization (a....
Upcoming SlideShare
Loading in …5
×

Siks December 2008 History Of Knowledge Representation

1,731 views

Published on

Published in: Education, Technology
0 Comments
0 Likes
Statistics
Notes
  • Be the first to comment

  • Be the first to like this

No Downloads
Views
Total views
1,731
On SlideShare
0
From Embeds
0
Number of Embeds
4
Actions
Shares
0
Downloads
37
Comments
0
Likes
0
Embeds 0
No embeds

No notes for slide
  • Siks December 2008 History Of Knowledge Representation

    1. 1. Rinke Hoekstra<br />History of Knowledge Representation<br />10-12-2008<br />SIKS Course - Knowledge Modelling<br />1<br />
    2. 2. Caveat Emptor<br />About me…<br />Knowledge Engineering<br />Ontologies<br />Web Ontology Language (OWL 2)<br />Dissertation<br />Ontology Representation: Design Patterns and Ontologies that Make Sense (to be published spring 2009, I hope)<br />10-12-2008<br />SIKS Course - Knowledge Modelling<br />2<br />
    3. 3. Overview<br />In the beginning… (400 BC – 1900s)<br />Scruffies vs. Neats (1970-ies) <br />The Dark Ages (1980-ies)<br />Engineering Revival (1990-ies)<br />The ‘O’ Word (1995 onwards)<br />10-12-2008<br />SIKS Course - Knowledge Modelling<br />3<br />
    4. 4. In the beginning…<br />400BC – 1900s<br />10-12-2008<br />SIKS Course - Knowledge Modelling<br />4<br />
    5. 5. Aristotle (384 BC – 322 BC)<br />Dialectics <br />reductio ad absurdum<br />Deduction<br />premises  conclusion (Plato)<br />Syllogisms<br />Standard logic until the 19th century<br />Categories<br />10-12-2008<br />SIKS Course - Knowledge Modelling<br />5<br />
    6. 6. Syllogisms<br />Example<br />Major premise All mortal things die<br />Minor premise All men are mortal things<br />Conclusion All men die <br />Forms<br />Names<br />Barbara (AAA), Celarent (EAE), …<br />10-12-2008<br />SIKS Course - Knowledge Modelling<br />6<br />
    7. 7. Aristotle’s Categories<br />Substance<br />primary vs. secondary<br />Quantity<br />extension<br />Quality<br />nature<br />Relation<br />Place<br />position relative to environment<br />Time<br />pos. relative to events<br />Position<br />condition of rest (action)<br />State<br />condition of rest (affection)<br />Action<br />production of change<br />Affection<br />reception of change<br />10-12-2008<br />SIKS Course - Knowledge Modelling<br />7<br />
    8. 8. Porphyry of Tyre (233–c. 309)<br />10-12-2008<br />SIKS Course - Knowledge Modelling<br />8<br />
    9. 9. Brentano (1838-1917)<br />10-12-2008<br />SIKS Course - Knowledge Modelling<br />9<br />
    10. 10. Ramon Llull (1232 – 1315)<br />Mechanical aids to reasoning<br />10-12-2008<br />SIKS Course - Knowledge Modelling<br />10<br />
    11. 11. Scientific Revolution (17th and 18th century)<br />Dualism<br />René Descartes (1596 – 1650)<br />Body as machine &lt;-&gt; Mind<br />Empiricism<br />John Locke (1632 – 1704)<br />Royal Society<br />Mercantilism<br />Engineering<br />Christiaan Huygens (1629 – 1695)<br />Blaise Pascal (1623 – 1662)<br />10-12-2008<br />11<br />
    12. 12. John Wilkins (1614 – 1672)<br />Universal Character<br />Replace latin<br />(Metric system)<br />Tree with 3 layers<br />10-12-2008<br />SIKS Course - Knowledge Modelling<br />12<br />
    13. 13. Gottfried Wilhelm Leibniz (1646 – 1716)<br />CharacteristicaUniversalis<br />“Once the characteristic numbers of most notions are determined, the human race will have a new kind of tool, a tool that will increase the power of the mind much more than optical lenses helped our eyes, a tool that will be as far superior to microscopes or telescopes as reason is to vision.”<br />(Leibniz, Philosophical Essays)<br />10-12-2008<br />SIKS Course - Knowledge Modelling<br />13<br />
    14. 14. Calculators<br />Pascaline<br />Addition<br />Substraction<br />Stepped Reckoner<br />Multiplication<br />Division<br />Binary System<br />… but Leibniz wanted more<br />Calculus Ratiocinator<br />10-12-2008<br />SIKS Course - Knowledge Modelling<br />14<br />
    15. 15. Another Leibniz Quote<br />&quot;If controversies were to arise, there would be no more need of disputation between two philosophers than between two accountants. For it would suffice to take their pencils in their hands, and say to each other: Let us calculate.”<br />Leibniz, Dissertio de Arte Combinatoria, 1666<br />10-12-2008<br />SIKS Course - Knowledge Modelling<br />15<br />
    16. 16. Linnaeus (1707-1778) –SystemaNaturae<br />10-12-2008<br />SIKS Course - Knowledge Modelling<br />16<br />
    17. 17. … so, what’s new?<br />Syllogisms<br />Rules of valid reasoning<br />Reasoning as Calculation<br />Symbol Manipulation<br />Categories<br />Top-down categories of thought<br />Universal Character/SystemaNaturae<br />Bottom-up inventory of phenomena in reality<br />10-12-2008<br />SIKS Course - Knowledge Modelling<br />17<br />
    18. 18. GottlobFrege (1884 – 1924)<br />Logic<br />Study of correct reasoning<br />Arithmetics and Mathematics<br />Begriffschrift<br />Formal Language (of Meaning)<br />Axiomatic Predicate Logic<br />Variables, Functions, Quantifiers<br />10-12-2008<br />SIKS Course - Knowledge Modelling<br />18<br />
    19. 19. Computers<br />Algorithms<br />Alan Turing (1912 – 1954)<br />Processor/Memory Architecture<br />Neumann JánosLajos(1903 – 1957)<br />Automatic Theorem Proving<br />Resolution<br />Artificial Intelligence! <br />But…<br />10-12-2008<br />SIKS Course - Knowledge Modelling<br />19<br />
    20. 20. Theorem Proving<br />``… great theorem proving controversy of the late sixties …’’ (Newell, 1982)<br />Problematic<br />No human scale <br />No organisation<br />No procedures<br />Small, theoretically hard problems<br />10-12-2008<br />SIKS Course - Knowledge Modelling<br />20<br />
    21. 21. Scruffies vs. Neats<br />1970ies<br />10-12-2008<br />SIKS Course - Knowledge Modelling<br />21<br />
    22. 22. Two Schools (1970ies and onwards)<br />Philosophy (Neats)<br />Clean, uniform language<br />Knowledge derives from small set of ‘elegant’ first principles<br />Theoretical understanding of reality<br />Cognitive Psychology (Scruffies)<br />Cognitively plausible language<br />Knowledge is what’s in our heads<br />Human intelligence and behaviour<br />10-12-2008<br />SIKS Course - Knowledge Modelling<br />22<br />
    23. 23. Artificial Intelligence<br />“. . . an entity is intelligent if it has an adequate modelof the world […], if it is clever enough to answer a wide variety of questions on the basis of this model, if it can get additional informationfrom the external world when required, and can perform such tasks in the external world as its goals demand and its physical abilities permit.” <br />(McCarthy and Hayes, 1969, p.4)<br />Frame Problem! <br />10-12-2008<br />SIKS Course - Knowledge Modelling<br />23<br />
    24. 24. Epistemic and Heuristic adequacy<br />McCarthy & Hayes:<br />Representation vs. Mechanism<br />Epistemic Adequacy<br />Correct representation<br />Heuristic Adequacy<br />Correct reasoning<br />10-12-2008<br />SIKS Course - Knowledge Modelling<br />24<br />
    25. 25. Heuristic vs. Epistemic views in Psychology <br />Knowledge is about the how<br />Problem Solving<br /> Production Systems<br />Knowledge is about the what<br />Natural Language<br />Memory<br /> Semantic Networks<br />10-12-2008<br />SIKS Course - Knowledge Modelling<br />25<br />
    26. 26. Information Processing System (IPS)<br />Computer as metaphor of the mind<br />“the human operates as an <br />information processing machine’’<br /> Newell & Simon, 1972<br />10-12-2008<br />SIKS Course - Knowledge Modelling<br />26<br />
    27. 27. Production Systems (1)<br />Processor<br />Interpreter<br />Elementary Information Processes (EIP)<br />Sequence of EIPs a function of symbols in memory<br />Production Rules (Emil Post, 1943)<br />if … then …<br />Rule ‘fires’ if interpreter finds a match between condition and symbols in memory<br />Sequential ≠ material implication<br />10-12-2008<br />SIKS Course - Knowledge Modelling<br />27<br />
    28. 28. Production Systems (2)<br />Adequacy?<br />Correspondence to human reasoning<br />Not ‘clean’ or ‘logical’<br />Escape limitations of theorem provers<br />Local, rational control of problem solving<br />Easily modifiable<br />Drawback: Natural language?<br />10-12-2008<br />SIKS Course - Knowledge Modelling<br />28<br />
    29. 29. Semantic Networks (1)<br />Natural Language<br />Ground lexical terms in a model of reality<br />Semantic Memory<br />M. Ross Quillian (1966)<br />Associative Memory<br />Semantic Networks<br />Graph Based<br />Nodes, planes and pointers<br />subclass, modification, disjunction, conjunction, subject/object<br />10-12-2008<br />SIKS Course - Knowledge Modelling<br />29<br />
    30. 30. Semantic Networks (2)<br />10-12-2008<br />SIKS Course - Knowledge Modelling<br />30<br />
    31. 31. Semantic Networks (3)<br />Adequacy?<br />Correspondence to human memory<br />Response time<br />Property inheritance<br />Extensions<br />Named Attributes (type/token)<br />Concepts vs. Examples (instances)<br />Jaime Carbonell, 1970<br />Sprawl of variants <br />10-12-2008<br />SIKS Course - Knowledge Modelling<br />31<br />
    32. 32. Frames (1)<br />Criticism from Cognitive Science<br />Frames, Marvin Minsky (1975)<br />Scripts, Roger Schank (1975)<br />Frames<br />Larger `chunks’ of thought<br />Situations (akin to planes)<br />Default values<br />10-12-2008<br />SIKS Course - Knowledge Modelling<br />32<br />
    33. 33. Frames (2)<br />Frame system<br />Related frames that share the same terminals<br />… different viewpoints on the same situation<br />Knowledge Reuse<br />Information Retrieval Network<br />Standard matching procedure<br />Fixed perspective: <br />situations, objects, processes <br />(object-oriented design)<br />10-12-2008<br />SIKS Course - Knowledge Modelling<br />33<br />
    34. 34. Semantic Networks (3)<br />Technical problems<br />Weak inference (inheritance)<br />Unclear semantics<br />“What’s in a link?”, Bill Woods (1975)<br />“What IS-A is and isn’t”, Ron Brachman (1983)<br />Consider the semantics of the representation itself<br />10-12-2008<br />SIKS Course - Knowledge Modelling<br />34<br />
    35. 35. Frame (like) Languages<br />Emphasis<br />Interrelated, internallystructuredconcepts<br />Knowledge Representation Language (KRL)<br />Bobrow and Winograd (1976)<br />Structured InheritanceNetworks<br />Ron Brachman (1979)<br />10-12-2008<br />SIKS Course - Knowledge Modelling<br />35<br />
    36. 36. Knowledge Representation Language (KRL)<br />Known entity: prototype<br />Description by reusable descriptors<br />Descriptions by comparison to prototype + extension<br />Modes of description:<br />membership, relationship, role (object/event)<br />Reasoning:<br />Process of recognition, procedural attachments<br />Inference mechanism determines meaning<br />10-12-2008<br />SIKS Course - Knowledge Modelling<br />36<br />
    37. 37. SI Networks<br />KL-ONE (Brachman, 1979; Brachman & Schmolze, 1985)<br />Descriptions<br />Role/Filler Descriptions<br />Structural Descriptions<br />Interpretive Attachments<br />Role modality types:<br />inherent, derivable, obligatory<br />10-12-2008<br />SIKS Course - Knowledge Modelling<br />37<br />
    38. 38. SI-Network of an Arch<br />10-12-2008<br />SIKS Course - Knowledge Modelling<br />38<br />
    39. 39. Epistemological Status<br />Cognitive plausibility Epist. Status<br />Relation to reality?<br />Relation to representation language?<br />10-12-2008<br />SIKS Course - Knowledge Modelling<br />39<br />
    40. 40. The Knowledge Level (Allen Newell, 1982)<br />“… the crux for AI is that no one has been able to formulate in a reasonable way the problem of finding the good representation, so that it can be tackled by an AI system” <br />(Newell, 1982, p.3)<br />Computer System Level<br />Medium<br />System<br />Processing Components<br />Composition Guidelines<br />Behavior <br />Independent, but reducible to lower level<br />10-12-2008<br />SIKS Course - Knowledge Modelling<br />40<br />
    41. 41. The Knowledge Level (2)<br />10-12-2008<br />SIKS Course - Knowledge Modelling<br />41<br />“There exists a distinct computer systems level, lying immediately above the symbol level, which is characterised by knowledge as the medium and the principle of rationality as the law of behaviour”<br />(Newell, 1982, p. 99)<br />
    42. 42. The Knowledge Level (3)<br />Not a stance<br />viz. the intentional stance(Dennett, 1987)<br />No representation at knowledge level<br />(concepts, tasks, goals)<br />Knowledge level = knowledge itself!<br />Representation always at the symbol level<br />Knowledge representation<br />Representation of knowledge, not reality<br />10-12-2008<br />SIKS Course - Knowledge Modelling<br />42<br />
    43. 43. Brachman’s Triangle Extended (Hoekstra, 2009)<br />10-12-2008<br />SIKS Course - Knowledge Modelling<br />43<br />
    44. 44. Representation and Language<br />Brachman’s levels in Semantic Nets<br />Primitives of KR languages<br />Requirements<br />neutrality, adequacy, well-defined semantics<br />10-12-2008<br />SIKS Course - Knowledge Modelling<br />44<br />
    45. 45. Epistemological Level<br />Missing level<br />Knowledge-structuring primitives<br />“The formal structure of conceptual units and their interrelationships as conceptual units (independent of any knowledge expressed therein) forms what could be called an epistemology.”<br />(Brachman, 1979, p.30)<br />Two interpretations<br />Adequacy of Language for some level<br />Representation at a level<br />e.g. Logical primitives as concepts<br />10-12-2008<br />SIKS Course - Knowledge Modelling<br />45<br />
    46. 46. Optimism<br />Modern Knowledge Representation<br />Representation of expert knowledge<br />Performance over Plausibility<br />Modern Languages<br />Defined semantics<br />Clear epistemological status<br />10-12-2008<br />SIKS Course - Knowledge Modelling<br />46<br />
    47. 47. The Dark Ages<br />1980ies<br />10-12-2008<br />SIKS Course - Knowledge Modelling<br />47<br />
    48. 48. Practical Applications (1980s)<br />Expert Systems<br />Production Rules<br />Rules of thumb<br />Relatively clear status<br />Memory in PSI of secondary importance<br />Severe problems<br />Scalability<br />Reusability<br />10-12-2008<br />SIKS Course - Knowledge Modelling<br />48<br />
    49. 49. MYCIN and GUIDON (William Clancey, 1983)<br />MYCIN: medical diagnosis<br />GUIDON: medical tutoring<br />“transfer back” expert knowledge<br />Problematic<br />No information about how the rule-base was structured: design knowledge<br />“Compiled Knowledge”<br />10-12-2008<br />SIKS Course - Knowledge Modelling<br />49<br />
    50. 50. Role of Knowledge in Problem Solving<br />10-12-2008<br />SIKS Course - Knowledge Modelling<br />50<br />
    51. 51. Knowledge Types<br />Order of rules: problem solving strategy<br />Structure in knowledge<br />Common causes before unusual ones<br />Justification: domain theory<br />Identification rules<br />Causal rules<br />World fact rules <br />Domain fact rules<br />10-12-2008<br />SIKS Course - Knowledge Modelling<br />51<br />
    52. 52. Convergence?<br />Heuristic vs. Epistemological Adequacy<br />Two approaches<br />Different formalisms<br />Same types of knowledge<br />Two solutions<br />Components (Clancey)<br />Knowledge Structuring (Brachman)<br />10-12-2008<br />SIKS Course - Knowledge Modelling<br />52<br />
    53. 53. Problems<br />Knowledge Acquisition Bottleneck (Feigenbaum, 1980)<br />The difficulty to correctly extract relevant knowledge from an expert into a knowledge base<br />Interaction Problem (Bylander and Chandrasekaran, 1987)<br />Different types of knowledge cannot be cleanly separated<br />Problem for reuse<br />10-12-2008<br />SIKS Course - Knowledge Modelling<br />53<br />
    54. 54. ENGINEERING REVIVAL<br />1990s<br />10-12-2008<br />SIKS Course - Knowledge Modelling<br />54<br />
    55. 55. Knowledge Acquisition<br />Ensure<br />Quality<br />Reuse<br />Ad hoc Methodologies<br />Engineering<br />Knowledge modelling vs. extraction<br />Implementation guided by Specification<br />10-12-2008<br />SIKS Course - Knowledge Modelling<br />55<br />
    56. 56. CommonKADS(Wielinga et al., 1992, van Heijst et al., 1997)<br />Knowledge Level Model<br />Independent of KR language<br />Solution to the KA Bottleneck?<br />Limited Interaction Hypothesis<br />Solution to the Interaction Problem?<br />10-12-2008<br />SIKS Course - Knowledge Modelling<br />56<br />
    57. 57. Reuse<br />Role limiting<br />Direct reuse<br />Index symbol level representations<br />Detailed blueprints<br />Skeletal Models<br />Reuse of `understanding’<br />Knowledge-level ‘sketches’<br />Library of reusable knowledge components<br />10-12-2008<br />SIKS Course - Knowledge Modelling<br />57<br />
    58. 58. Knowledge Types (1)<br />Control Knowledge<br />Task Knowledge<br />Inference Knowledge<br />Problem Solving Methods (Breuker & van de Velde, 1994)<br />10-12-2008<br />SIKS Course - Knowledge Modelling<br />58<br />
    59. 59. Knowledge Types (2)<br />Domain Knowledge<br />Index PSM’s for reuse  Epistemology<br />Generic domain theory<br />What an expert system ‘knows’ about:<br />ONTOLOGY<br />10-12-2008<br />SIKS Course - Knowledge Modelling<br />59<br />
    60. 60. Functional Perspective (Hector Levesque, 1984)<br />Descend to the Symbol Level?<br />Knowledge Base<br />Abstract datatype<br />Competencies<br />Set of TELL/ASK queries<br />Capabilities of KB<br />Function of queries/answers, assertions<br />10-12-2008<br />SIKS Course - Knowledge Modelling<br />60<br />
    61. 61. Knowledge Based Systems<br />Architecture<br />Specialised KR languages<br />Specialised Services<br />Performance guarantees<br />Domain Theory <br />Identification, Classification<br />KL-ONE like languages… <br />Control Knowledge<br />Rules…<br />10-12-2008<br />SIKS Course - Knowledge Modelling<br />61<br />
    62. 62. The return of logic (Levesque & Brachman, 1987)<br />Classification as logical inference<br />Exact semantics<br />Trade-off<br />Expressive power<br />Computational efficiency<br />Restricted Language Thesis<br />“… general purpose knowledge representation systems should restrict their languages by omitting constructs which require non-polynomial (or otherwise unacceptably long) worst-case response times for correct classification of concepts.” (Doyle & Patil, 1991)<br />10-12-2008<br />SIKS Course - Knowledge Modelling<br />62<br />
    63. 63. Description Logics (Baader & Hollunder, 1991)<br />KL-One, NIKL, KL-Two, LOOM, FL, KANDOR, KRYPTON, CLASSIC …<br />Quest<br />Expressive<br />Sound & Complete<br />Decidable<br />KRIS, SHIQ, SHOIN, SROIQ, …<br />10-12-2008<br />SIKS Course - Knowledge Modelling<br />63<br />
    64. 64. … and the rest?<br />Domain Theory <br />Causal Knowledge<br />Naïve Physics<br />Qualitative Reasoning (J. de Kleer, K.D. Forbus, B. Bredeweg, …)<br />Strategic Knowledge<br />Logic-based approaches<br />Prolog, Datalog, etc..<br />… no principled effort.<br />10-12-2008<br />SIKS Course - Knowledge Modelling<br />64<br />
    65. 65. The ‘O’ Word<br />1995 and onwards<br />10-12-2008<br />SIKS Course - Knowledge Modelling<br />65<br />Oh no! <br />Not that again!<br />
    66. 66. Pop Quiz<br />What is an ontology?<br />10-12-2008<br />SIKS Course - Knowledge Modelling<br />66<br />
    67. 67. Ontology<br />“Ontology or the science of something and of nothing, of being and not-being of the thing and the mode of the thing, of substance and accident”<br />G.W. Leibniz<br />“… ontology, the science, namely, which is concerned with the more general properties of all things.”<br />Immanuel Kant<br />The nature of being<br />Aristotle’s categories<br />10-12-2008<br />SIKS Course - Knowledge Modelling<br />67<br />
    68. 68. Knowledge Representation (Davis, Shrobe, Szolovits, 1993)<br />Surrogate<br />Set of ontological commitments<br />through language and domain theory<br />Fragmentary theory of intelligent reasoning<br />sanctions heuristic adequacy<br />Medium for pragm. efficient computation<br />way of formulating problems (Newell)<br />Medium of human expression<br />``Universal Character’’(Wilkins, Leibniz, … and Stefik, 1986)<br />10-12-2008<br />SIKS Course - Knowledge Modelling<br />68<br />
    69. 69. Ontology Definitions<br />Knowledge Management<br />An explicit (knowledge level) specification of a conceptualization (a.o. Gruber, 1994)<br />Knowledge Representation<br />An explicit (symbol level) specification of a conceptualisation<br />Philosophy<br />A formal specification of an ontological theory<br />10-12-2008<br />SIKS Course - Knowledge Modelling<br />69<br />
    70. 70. The END<br />10-12-2008<br />SIKS Course - Knowledge Modelling<br />70<br />

    ×