CS661 Artifical Intelligence
Lecture 6 - Propositional Logic
- Knowledge based agents (RN s6.1)
- more efficient than search since can reason
- KB collection of facts, usually stored as sentences or tables
- Tell, Ask primitives, how does Ask know answers? logical inference
- Example - Wumpus world (RN s6.2)
- Logic (RN s6.3)
- syntax (grammar)
- semantics (meaning)
- correct syntax but not semantics: The smart guitar ran down the cloud.
- inference procedure (proof method)
- Mechanization of logic
- Propositional logic
- vs. predicate logic, probabilistic logic, fuzzy logic. etc.
- syntax elements: constants, dummy variables, not and or if iff, parentheses
- can only check veracity of proposition - tautology or satisfiability
- syntax
sentence ::= atom | complex_sentence
atom ::= true, false, P, Q, ... (variables)
complex_sentence ::= (sentence) | - sentence | sentence connective sentence
connective ::= and or if iff
examples of wffs and non-wffs
semantics
- true = anything which exists in the world = environment
- false = opposite of true
- P, Q any statement about environment that has true/false answer
- and, or, if, iff (xor) meanings, show if, iff, etc in terms of and/or/not
- 4 possible 1-input/1-output functions, only not nontrivial, truth tables
- 16 possible 2-input/2-output functions (connectives)
- sufficiency - not+and+or, not+and, not+or, nand, nor
- parentheses and conventions
- cnf, dnf, Horn forms (expert systems)
systems equivalent to logic
- boolean variables in GF2 (and thus computer software)
- set theory and Venn diagrams
- electronics, eg. relays (and thus computer hardware)
Wumpus world (RN s6.2)
Decidability
- truth tables
- complexity - 3SAT is NP-complete
Simplification inference rules