CPECSC 580 Knowledge Management Dr Franz J Kurfess

CPE/CSC 580: Knowledge Management Dr. Franz J. Kurfess Computer Science Department Cal Poly © 2001 -2005 Franz J. Kurfess Knowledge Processing 1

Course Overview u u Introduction Knowledge Processing u u Knowledge Organization u u u Classification, Categorization Ontologies, Taxonomies, Thesauri Knowledge Retrieval u u u Knowledge Acquisition, Representation and Manipulation Information Retrieval Knowledge Navigation Knowledge Presentation u Knowledge Visualization © 2001 -2005 Franz J. Kurfess u Knowledge Capture, Transfer, and Distribution u Usage u of Knowledge Access Patterns, User Feedback u Knowledge Techniques u Exchange Management Topic Maps, Agents u Knowledge Management Tools u Knowledge Management in Organizations Knowledge Processing 2

Overview Knowledge Processing u Motivation u Knowledge u Objectives u u Chapter u u Introduction Knowledge Processing as Core AI Paradigm Relationship to KM Terminology u Knowledge u u Acquisition Knowledge Elicitation Machine Learning © 2001 -2005 Franz J. Kurfess u u Logic Rules Semantic Networks Frames, Scripts u Knowledge u u Representation Manipulation Reasoning KQML u Important Concepts and Terms u Chapter Summary Knowledge Processing 3

Motivation u the representation and manipulation of knowledge has been essential for the development of humanity as we know it u the use of formal methods and support from machines can improve our knowledge representation and reasoning abilities u intelligent reasoning is a very complex phenomenon, and may have to be described in a variety of ways u a basic understanding of knowledge representation and reasoning is important for the organization and management of knowledge © 2001 -2005 Franz J. Kurfess Knowledge Processing 7

Objectives u be familiar with the important aspects of commonly used knowledge representation and reasoning methods u understand different roles and perspectives of knowledge representation and reasoning methods u examine the suitability of knowledge representations for specific tasks u evaluate the representation methods and reasoning mechanisms employed in computer-based systems © 2001 -2005 Franz J. Kurfess Knowledge Processing 8

Chapter Introduction u Knowledge Processing as Core AI Paradigm u Relationship to KM u Terminology © 2001 -2005 Franz J. Kurfess Knowledge Processing 10

Knowledge u knowledge characteristics u meaningful only with respect to humans u context-sensitive u may be elaborate u may be explicit or tacit v explicit knowledge consists of documented facts v v v frequently objective can be “spelled out” tacit knowledge is in people’s heads v v frequently subjective surfaces through interaction © 2001 -2005 Franz J. Kurfess [Knowledge Ability 1998] Knowledge Processing 18

Knowledge Processes Chaotic knowledge processes Human knowledge and networking Information databases and technical networking Systematic information and knowledge processes © 2001 -2005 Franz J. Kurfess [Skyrme 1998] Knowledge Processing 19

Knowledge Cycles Collect Codify Identify Embed Product/ Process Diffuse © 2001 -2005 Franz J. Kurfess Create Classify Knowledge Repository Use/Exploit Access [Skyrme 1998] Organize/ Store Share/ Disseminate Knowledge Processing 20

Knowledge Representation u Types of Knowledge u Factual Knowledge u Subjective Knowledge u Heuristic Knowledge u Deep and Shallow Knowledge u Knowledge Representation Methods u Rules, Frames, Semantic Networks u Blackboard Representations u Object-based Representations u Case-Based Reasoning u Knowledge Representation Tools © 2001 -2005 Franz J. Kurfess Knowledge Processing 21

Roles of Knowledge Representation u Surrogate u Ontological Commitments u Fragmentary Theory of Intelligent Reasoning u Medium for Computation u Medium for Human Expression © 2001 -2005 Franz J. Kurfess [Davis, Shrobe, Szolovits, 1993] Knowledge Processing 22

KR as Surrogate ua substitute for the thing itself u enables an entity to determine consequences by thinking rather than acting u reasoning about the world through operations on the representation u reasoning or thinking are inherently internal processes u the objects of reasoning are mostly external entities (“things”) u some objects of reasoning are internal, e. g. concepts, feelings, . . . © 2001 -2005 Franz J. Kurfess [Davis, Shrobe, Szolovits, 1993] Knowledge Processing 23

Surrogate Aspects u Identity u correspondence between the surrogate and the intended referent in the real world u Fidelity u Incompleteness u Incorrectness u Adequacy Task v User v © 2001 -2005 Franz J. Kurfess [Davis, Shrobe, Szolovits, 1993] Knowledge Processing 24

Surrogate Consequences u perfect representation is impossible u the only completely accurate representation of an object is the object itself u incorrect reasoning is inevitable u if there are some flaws in the world model, even a perfectly sound reasoning mechanism will come to incorrect conclusions © 2001 -2005 Franz J. Kurfess [Davis, Shrobe, Szolovits, 1993] Knowledge Processing 25

Ontological Commitments u terms used to represent the world u by selecting a representation a decision is made about how and what to see in the world u like a set of glasses that offer a sharp focus on part of the world, at the expense of blurring other parts u necessary because of the inevitable imperfections of representations u useful to concentrate on relevant aspects u pragmatic because of feasibility constraints © 2001 -2005 Franz J. Kurfess [Davis, Shrobe, Szolovits, 1993] Knowledge Processing 26

Ontological Commitments Examples u logic u views the world in terms of individual entities and relationships between the entities u rules u entities and their relationships expressed through rules u frames u prototypical u semantic u entities objects nets and relationships © 2001 -2005 Franz J. Kurfess [Davis, Shrobe, Szolovits, 1993] Knowledge Processing 27

KR and Reasoning ua knowledge representation indicates an initial conception of intelligent inference u often reasoning methods are associated with representation technique first order predicate logic and deduction v rules and modus ponens v u the association is often implicit u the underlying inference theory is fragmentary the representation covers only parts of the association v intelligent reasoning is a complex and multi-faceted phenomenon v © 2001 -2005 Franz J. Kurfess [Davis, Shrobe, Szolovits, 1993] Knowledge Processing 28

KR for Reasoning ua representation suggests answers to fundamental questions concerning reasoning: u What v implied reasoning method u What v can possibly be inferred from what we know? possible conclusions u What v does it mean to reason intelligently? should be inferred from what we know? recommended conclusions © 2001 -2005 Franz J. Kurfess [Davis, Shrobe, Szolovits, 1993] Knowledge Processing 29

KR and Computation u for our purposes, reasoning is a computational process u machines are used as reasoning tools u without efficient ways of implementing such computational process, it is practically useless u e. g. Turing machine u most representation and reasoning mechanisms are modified for efficient computation u e. g. Prolog vs. predicate logic © 2001 -2005 Franz J. Kurfess [Davis, Shrobe, Szolovits, 1993] Knowledge Processing 30

Computational Medium u computational environment for the reasoning process u reasonably efficient u organization of knowledge so that reasoning is facilitated © 2001 -2005 Franz J. Kurfess Knowledge Processing 31

KR for Human Expression ua language that can be used by humans to make statements about the world u expression v of knowledge expressiveness, generality, preciseness u communication of knowledge among humans v between humans and machines v among machines v © 2001 -2005 Franz J. Kurfess [Davis, Shrobe, Szolovits, 1993] Knowledge Processing 32

Knowledge Acquisition u Knowledge Elicitation u Machine Learning © 2001 -2005 Franz J. Kurfess Knowledge Processing 33

Acquisition of Knowledge u Published Sources u Physical Media u Digital Media u People as Sources u Interviews u Questionnaires u Formal Techniques u Observation Techniques u Knowledge Acquisition Tools © 2001 -2005 Franz J. Kurfess Knowledge Processing 34

Knowledge Elicitation u knowledge is already present in humans, but needs to be converted into a form suitable for computer use u requires the collaboration between a domain expert and a knowledge engineer u domain expert has the domain knowledge, but not necessarily the skills to convert it into computer-usable form u knowledge engineer assists with this conversion u this can be a very lengthy, cumbersome and error-prone process © 2001 -2005 Franz J. Kurfess Knowledge Processing 35

Machine Learning u extraction of higher-level information from raw data u based on statistical methods u results are not necessarily in a format that is easy for humans to use u the organization of the gained knowledge is often far from intuitive for humans u examples u decision trees u rule extraction from neural networks © 2001 -2005 Franz J. Kurfess Knowledge Processing 36

Knowledge Fusion u integration of human-generated and machinegenerated knowledge u sometimes also used to indicate the integration of knowledge from different sources, or in different formats u can be both conceptually and technically very difficult u different “spirit” of the knowledge representation used u different terminology u different categorization criteria u different representation and processing mechanisms © 2001 -2005 Franz J. Kurfess Knowledge Processing 37

Knowledge Representation Mechanisms u Logic u Rules u Semantic Networks u Frames, Scripts © 2001 -2005 Franz J. Kurfess Knowledge Processing 38

Logic u syntax: well-formed formula ua formula or sentence often expresses a fact or a statement u semantics: interpretation of the formula u “meaning” is associated with formulae u often compositional semantics u axioms as basic assumptions u generally accepted within the domain u inference rules for deriving new formulae from existing ones © 2001 -2005 Franz J. Kurfess Knowledge Processing 39

KR Roles and Logic u surrogate u very expressive, not very suitable for many types of knowledge u ontological u objects, commitments relationships, terms, logic operators u fragmentary u deduction, u medium u yes, other logical calculi for computation but not very efficient u medium u only theory of intelligent reasoning for human expression for experts © 2001 -2005 Franz J. Kurfess Knowledge Processing 40

Rules u syntax: if … then … u semantics: interpretation of rules u usually u initial reasonably understandable rules and facts u often capture basic assumptions and provide initial conditions u generation of new facts, application to existing rules u forward reasoning: starting from known facts u backward reasoning: starting from a hypothesis © 2001 -2005 Franz J. Kurfess Knowledge Processing 41

KR Roles and Rules u surrogate u reasonably expressive, suitable for some types of knowledge u ontological u objects, commitments rules, facts u fragmentary theory of intelligent reasoning u modus ponens, matching, sometimes augmented by probabilistic mechanisms u medium for computation u reasonably u medium efficient for human expression mainly for experts © 2001 -2005 u Franz J. Kurfess Knowledge Processing 42

Semantic Networks u syntax: graphs, possibly with some restrictions and enhancements u semantics: interpretation of the graphs u initial state of the graph u propagation of activity, inferences based on link types © 2001 -2005 Franz J. Kurfess Knowledge Processing 43

KR Roles and Semantic Nets u surrogate u limited to reasonably expressiveness, suitable for some types of knowledge u ontological u nodes commitments (objects, concepts), links (relations) u fragmentary theory of intelligent reasoning u conclusions based on properties of objects and their relationships with other objects u medium for computation u reasonably u medium efficient for some types of reasoning for human expression easy to visualize © 2001 -2005 u Franz J. Kurfess Knowledge Processing 44

Frames, Scripts u syntax: templates with slots and fillers u semantics: interpretation of the slots/filler values u initial values for slots in frames u complex matching of related frames © 2001 -2005 Franz J. Kurfess Knowledge Processing 45

KR Roles and Frames u surrogate u suitable for well-structured knowledge u ontological commitments u templates, situations, properties, methods u fragmentary u conclusions u medium u ok theory of intelligent reasoning are based on relationships between frames for computation for some problem types u medium u ok, for human expression but sometimes too formulaic © 2001 -2005 Franz J. Kurfess Knowledge Processing 46

Knowledge Manipulation u Reasoning u KQML © 2001 -2005 Franz J. Kurfess Knowledge Processing 47

Reasoning u generation of new knowledge items from existing ones u frequently identified with logical reasoning u strong formal foundation u very restricted methods for generating conclusions u sometimes expanded to capture various ways to draw conclusions based on methods employed by humans u requires a formal specification or implementation to be used with computers © 2001 -2005 Franz J. Kurfess Knowledge Processing 48

KQML u stands for Knowledge Query and Manipulation Language u language and protocol for exchanging information and knowledge © 2001 -2005 Franz J. Kurfess Knowledge Processing 49

KQML Performatives u basic u query performatives evaluate, ask-if, ask-about, ask-one, ask-all u multi-response u stream-about, stream-all u response u informational performatives tell, achieve, deny, untell, unachieve u generator u performatives reply, sorry u generic u query performatives standby, ready, next, rest, discard, generator u capability-definition u advertise, subscribe, monitor, import, export u networking u performatives register, unregister, forward, broadcast, route. © 2001 -2005 Franz J. Kurfess Knowledge Processing 50

KQML Example 1 u query u reply (ask-if : sender A : receiver B : language Prolog : ontology foo : reply-with id 1 : content ``bar(a, b)'' ) (sorry : sender B : receiver A : in-reply-to id 1 : reply-with id 2 ) agent A (: sender) is querying the agent B (: receiver), in Prolog (: language) about the truth status of ``bar(a, b)'' (: content) © 2001 -2005 Franz J. Kurfess Knowledge Processing 51

KQML Example 2 u query u reply (stream-about : language KIF : ontology motors `: replywith q 1 : content motor 1) (tell : language KIF : ontology motors : inreply-to q 1 : content (= (val (torque motor 1) (sim-time 5) (scalar 12 kgf)) (tell : language KIF : ontology structures : inreply-to q 1 : content (fastens frame 12 motor 1)) (eos : in-repl-to q 1) agent A asks agent B to tell all it knows about motor 1. B replys with a sequence of tells terminated with a sorry. © 2001 -2005 Franz J. Kurfess Knowledge Processing 52

KP/KM Activity u select a domain that requires significant human involvement for dealing with knowledge u identify at least two candidates for u knowledge representation u reasoning u evaluate u human v their suitability perspective understandable and usable for humans u computational v perspective storage, processing © 2001 -2005 Franz J. Kurfess Knowledge Processing 55

Important Concepts and Terms u u u u automated reasoning belief network cognitive science computer science deduction frame human problem solving inference intelligence knowledge acquisition knowledge representation linguistics logic machine learning u u u u © 2001 -2005 Franz J. Kurfess natural language ontology ontological commitment predicate logic probabilistic reasoning propositional logic psychology rational agent rationality reasoning rule-based system semantic network surrogate taxonomy Turing machine Knowledge Processing 56

Summary Knowledge Processing © 2001 -2005 Franz J. Kurfess Knowledge Processing 57

© 2001 -2005 Franz J. Kurfess Knowledge Processing 58