Symbolic Representation and Reasoning an Overview Stuart C. Shapiro Department of Computer Science and Engineering, Center for Multisource Information Fusion, and Center for Cognitive Science University at Buffalo, The State University of New York 201 Bell Hall, Buffalo, NY 14260-2000 shapiro@cse.buffalo.edu http://www.cse.buffalo.edu/~shapiro/ Introduction Knowledge Representation Reasoning Symbols Logics September, 2004 S. C. Shapiro 2 Knowledge Representation A subarea of Artificial Intelligence Concerned with understanding, designing, and implementing ways of representing information in computers So that programs can use this information to derive information that is implied by it, to converse with people in natural languages, to plan future activities, to solve problems in areas that normally require human expertise. September, 2004 S. C. Shapiro 3 Reasoning Deriving information that is implied by the information already present is a form of reasoning. Knowledge representation schemes are useless without the ability to reason with them. So, Knowledge Representation and Reasoning September, 2004 S. C. Shapiro 4 Knowledge vs. Belief Knowledge: Justified True Belief KR systems operate the same whether or not the information stored is justified or true. So, Belief Representation and Reasoning would be better. But we’ll stick with KR. September, 2004 S. C. Shapiro 5 What Is a Symbol? “A symbol token is a pattern that can be compared to some other symbol token and judged equal with it or different from it… Symbols may be formed into symbol structures by means of a set of relations… The `objects’ that symbols designate may include … objects in an external environment of sensible (readable) stimuli.” [Newell & Simon, Concise Encyclopedia of CS, 2004] September, 2004 S. C. Shapiro 6 What Is Logic? The study of correct reasoning. Not a particular KR language. There are many systems of logic. With slight abuse, we call a system of logic a logic. KR research may be seen as the search for the correct logic(s) to use in intelligent systems. September, 2004 S. C. Shapiro 7 Parts of Specifying a Logic Syntax Semantics Proof Theory September, 2004 S. C. Shapiro 8 Syntax The specification of a set of atomic symbols, and the grammatical rules for combining them into well-formed expressions (symbol-structures). September, 2004 S. C. Shapiro 9 Syntactic Expressions Atomic symbols Individual constants: Tom, Betty, white Variables: x, y, z Function symbols: motherOf Predicate symbols: Person, Elephant, Color Propositions: P, Q, BdT Terms Individual constants: Tom, Betty, white Variables : x, y, z Functional terms: motherOf(Fred) Well-formed formulas (wffs) Propositions (Proposition symbols) : P, Q, BdT Atomic formulas: Color(x, white), Duck(motherOf(Fred)) Non-atomic formulas: TdB Td Bp September, 2004 S. C. Shapiro 10 Semantics The specification of the meaning (designation) of the atomic symbols, and the rules for determining the meanings of the well-formed expressions from the meanings of their parts. September, 2004 S. C. Shapiro 11 Semantic Values Terms could denote Objects Categories of objects Properties… Wffs could denote Propositions Truth values September, 2004 S. C. Shapiro 12 Truth Values Could be 2, 3, 4, …, ∞ different truth values. Some truth values are “distinguished” Needn’t have anything to do with truth in the real world. By default, we’ll assume 2 truth values. Call distinguished one True (T) Call other False (F) September, 2004 S. C. Shapiro 13 Proof Theory The specification of a set of rules, which, given an initial collection of well-formed expressions, specify what other well-formed expressions can be added to the collection. September, 2004 S. C. Shapiro 14 Proof / Knowledge Base The collection could be A proof A knowledge base The initial collection could be Axioms Hypotheses Assumptions Domain facts & rules The added expressions could be Theorems Derived facts & rules September, 2004 S. C. Shapiro 15 Example Logic: Standard Propositional Logic Domain: CarPool World Atomic Proposition Symbols: BdT, TdB, Bd, Td, Bp, Tp Unary wff-forming connective: Binary wff-forming connectives: , , , September, 2004 S. C. Shapiro 16 Intended Interpretation (Intensional Semantics) BdT: Betty drives Tom TdB: Tom drives Betty Bd: Betty is the driver Td: Tom is the driver Bp: Betty is the passenger Tp: Tom is the passenger September, 2004 S. C. Shapiro 17 Extensional (Denotational) Semantics BdT TdB Bd T T T T T T T F T F T F F F F Td Bp Tp T T T T F F F F F F T T T T F F F F F F T F T F T F T F T F Bd Tp Td Td Td Td 5 of 26 = 64 possible situations September, 2004 S. C. Shapiro 18 Properties of Wffs Satisfiable T in some situation BdT T T T F F TdB T T F T F Bd T T T F F Td T F F T F Bp T F F T F Tp T F T F F Bd Tp T F T F F Td Td T T T T T Td Td F F F F F September, 2004 S. C. Shapiro 19 Properties of Wffs Contingent T in some, F in some BdT T T T F F TdB T T F T F Bd T T T F F Td T F F T F Bp T F F T F Tp T F T F F Bd Tp T F T F F Td Td T T T T T Td Td F F F F F September, 2004 S. C. Shapiro 20 Properties of Wffs Valid T in all situations BdT T T T F F TdB T T F T F Bd T T T F F Td T F F T F Bp T F F T F Tp T F T F F Bd Tp T F T F F Td Td T T T T T Td Td F F F F F September, 2004 S. C. Shapiro 21 Properties of Wffs Contradictory T in no situation BdT T T T F F TdB T T F T F Bd T T T F F Td T F F T F Bp T F F T F Tp T F T F F Bd Tp T F T F F Td Td T T T T T Td Td F F F F F September, 2004 S. C. Shapiro 22 Logical Implication P1, …, Pn logically imply Q P1, …, Pn |= Q In every situation that P1, …, Pn are True, so is Q. September, 2004 S. C. Shapiro 23 Example: CarPool World KB Let KBCPW = Bd Bp Td Tp BdT Bd Tp TdB Td Bp TdB BdT September, 2004 S. C. Shapiro 24 Extensional (Denotational) Semantics BdT TdB Bd T F T F T F Td Bp Tp F F T T T F Only 2 of the 64 situations where KBCPW are T So, e.g., KBCPW, BdT |= Bd Bp This is how a KB constrains a model to the domain we want. September, 2004 S. C. Shapiro 25 Proof Theory Some Rules of Inference PQ P Q Modus Ponens or Elimination Q Elimination PQ P P Q PQ Elimination September, 2004 PQ P Introduction S. C. Shapiro 26 Derivation from Assumptions Q is derivable from P1, …, Pn P1, …, Pn |- Q Starting from the collection P1, …, Pn, one can repeatedly apply rules of inference, and eventually get Q. September, 2004 S. C. Shapiro 27 Example: CarPool World Proof BdT Bd Tp BdT Bd Bp Bd Tp Bd Bp Bd Bp So, KBCPW, BdT |- Bd Bp September, 2004 S. C. Shapiro 28 Theoremhood If Q is derivable from no assumptions, |- Q We say that Q is provable, and that Q is a theorem. September, 2004 S. C. Shapiro 29 Deduction Theorem P1, …, Pn |= Q iff |= (P1 · · · Pn ) Q P1, …, Pn |- Q iff |- (P1 · · · Pn ) Q So theorem-proving is relevant to reasoning. September, 2004 S. C. Shapiro 30 Properties of Logics Soundness If |- P then |= P (If P is a provable, then P is valid.) Completeness If |= P then |- P (If P is valid, then P is a provable.) September, 2004 S. C. Shapiro 31 Soundness vs. Completeness Soundness is the essence of correct reasoning Completeness is less important because it doesn’t indicate how long it might take. September, 2004 S. C. Shapiro 32 Commutativity Diagram for Sound and Complete Logics |= (P1 · · · Pn ) Q P1, …, Pn |= Q soundness soundness completeness completeness |- (P1 · · · Pn ) Q P1, …, Pn |- Q So, whenever you want one, you can do another. September, 2004 S. C. Shapiro 33 Use of Commutativity Diagram Refutation proof techniques, such as resolution refutation or semantic tableaux, prove that there can be no situation in which P1, …, and Pn are True and Q is False. These are semantic proof techniques. September, 2004 S. C. Shapiro 34 Decision Procedure A procedure that is guaranteed to terminate and tell whether or not P is provable. September, 2004 S. C. Shapiro 35 Semidecision Procedure A procedure that, if P is a theorem is guaranteed to terminate and say so. Otherwise, it may not terminate. September, 2004 S. C. Shapiro 36 A Tour of Some Classes of Logics Propositional Logics Elementary Predicate Logics Full First-Order Logics September, 2004 S. C. Shapiro 37 Propositional Logics Smallest Unit: Proposition/Sentence propositional logics that are Sound Complete Have decision procedures September, 2004 S. C. Shapiro 38 What You Can Do with Propositional Logic • BettyDrivesTom TomDrivesBetty • BettyDrivesTom NearTomBetty • TomDrivesBetty NearTomBetty • NearTomBetty Can derive conclusions even though the “facts” aren’t entirely known. September, 2004 S. C. Shapiro 39 Elementary Predicate Logics Propositions plus Predicate (Relation) symbols, Individual terms, variables, quantifiers elementary predicate logics that are Sound Complete Have decision procedures September, 2004 S. C. Shapiro 40 What You Can Say with Elementary Predicate Logic • x[Elephant(x) HasA(x, trunk)] Can state generalities before all individuals are known. • x[Elephant(x) Color(x, white)] Can describe individuals Even when they are not specifically known. September, 2004 S. C. Shapiro 41 Full First-Order Logics Elementary predicate logic plus Function symbols/ functional terms full first-order logics that are Sound None are Complete Have decision procedures September, 2004 S. C. Shapiro 42 What You Can Say with Full First-Order Logic p[HasProp(0, p) x[HasProp(x, p) HasProp(x+1, p)] x HasProp(x, p)] Principle of induction. September, 2004 S. C. Shapiro 43 Example of Undecidability • Large KB about ducks, etc. • x[y (Duck(y) WalksLike(x,y)) y (Duck(y) TalksLike(x,y)) Duck(x)] • x Duck(motherOf(x)) Duck(x) • Duck(Fred)? • If Fred is not a duck, possible ∞ loop. September, 2004 S. C. Shapiro 44 Unsound Reasoning Induction From Raven(a) Black(a) Raven(b) Black(b) Raven(c) Black(c) Raven(d) Black(d) … Raven(n) Black(n) To x[Raven(x) Black(x)] September, 2004 S. C. Shapiro 45 Unsound Reasoning Abduction From x[Person(x) Injured(x) Bandaged(x)] Person(Tom) Bandaged(Tom) To Injured(Tom) September, 2004 S. C. Shapiro 46 What’s “First-Order” about First-Order Logics Can’t quantify over Function symbols Predicate symbols Propositions September, 2004 S. C. Shapiro 47 Examples of SNePS Reasoning Using a Logic Designed for KRR September, 2004 S. C. Shapiro 48 SNePS, A “Higher-Order” Logic : all(R)(Transitive(R) => (all(x,y,z)(R(x,y) and R(y,z) => R(x,z)))). : Bigger(elephants, lions). : Bigger(lions, mice). : Transitive(Bigger). : Bigger(elephants, mice)? Bigger(elephants,mice) Really a higher-order language for a first-order logic September, 2004 S. C. Shapiro 49 “Higher-Order” Example 2 : all(source)(Trusted(source) => all(p)(Says(source, p) => p)). : Trusted(Agent007). : Says(Agent007, Dangerous(Dr_No)). : Dangerous(Dr_No)? Dangerous(Dr_No) September, 2004 S. C. Shapiro 50 Designing New Connectives : andor(1,1){OnFloor(G2), OnFloor(G1), OnFloor(1), OnFloor(2)}. : OnFloor(G1). : OnFloor(?where)? ~OnFloor(G2) ~OnFloor(1) ~OnFloor(2) OnFloor(G1) September, 2004 S. C. Shapiro 51 Belief Change : andor(1,1){OnFloor(G2), OnFloor(G1), OnFloor(1), OnFloor(2)}. : {OnFloor(G2), OnFloor(G1)} => {Location(belowGround)}. : {OnFloor(1), OnFloor(2)} => {Location(aboveGround)}. : perform believe(OnFloor(G2)) : Location(?where)? Location(belowGround) : perform believe(OnFloor(2)) : Location(?where)? Location(aboveGround) September, 2004 S. C. Shapiro 52 Summary 1 Symbolic KRR uses logic. There are many logics. The question is which to use. September, 2004 S. C. Shapiro 53 Summary 2 A logic has a Syntax Semantics Proof Theory Logics may Be sound Be complete Have a decision procedure September, 2004 S. C. Shapiro 54 Summary 3 Logics provide non-atomic wffs That can describe situations Without knowing all specifics September, 2004 S. C. Shapiro 55 Summary 4 One can design and build Useful new logics And reasoning systems using them. September, 2004 S. C. Shapiro 56

Descargar
# Self-Aware Cassie - University at Buffalo