Knowledge Respresentation

Knowledge Respresentation

1. Representation Examples

   • First Order Logic
   • Logic Programs
   • Neural Network

2. General Representations

   • Logical Representations
     • A language with concrete rules
       • 1. No ambiguity in logic. 2. No ambiguity in processing and communication 3. Unlike Natural Languages (English)
     • Many ways to translate between languages
       • A single statement can be represented in different logics The same statement can be represented differently in the same logic
     • Expressiveness of logic
       • How much can we say in a logic
     • Logics are languages. Reasoning is a process
       • Logical reasoning is different from logical representation.
     • Logic is a good Representation
       • Fairly easy to do translations
       • Enables us to do logical reasoning
       • Basis for programming languages
         • Prolog uses logic programs (a Subset of FOL) Prolog is based on HOL
   • Description Logic
   • Propositional Logic
     • Syntax
       • Proposition
       • Connectives
       • Brackets
     • Semantics
       • Define how connectives affect truth
     • Propositional Logic Combines atoms
       • An atom does not contain propositional connectives
       • Have no Structure
         • eg. Today_Is_Wet
   • Predicate Logic
     • Allows us to talk about objects
       • True/False
         • Beyond True/False
           • Multi Valued Logic
             • More Than two truth values
             • Fuzzy Logic uses probability
               • values between [0,1]
           • Modal Logic
             • Model operators define mode for propositions
             • Epistemic Logic (Belief)
               • Necessarily Possibly
             • Temporal Logic (Time)
               • Always Eventually
       • Relations
       • Has Properties
     • Each atom is a predicate
       • eg. 1. First Order Logic (FOL) 2. Higher Order Logic (HOL)
       • First Order Logic
         • More expressive logic than propositional
         • Constants are objects
         • Predicates are properties and relations
         • Functions transform objects
         • Variables represent any object
         • Quantifiers qualify values of variables
       • HIgher Order Logic
         • More expressive than FOL
         • Functions and Predicates are also objects
         • Harder to reason with
   • Non-Logic Representation
     • Production Rules
       • Rule set of <Condition, Action> pairs
         • IF condition THEN Action
       • Requires some Match-Resolve-Act cycle
         • Match: Agent checks if the rule's condition holds
         • Resolve
           • Conflict Set
             • Multiple production rules may fire at once
           • Conflict Resolution
             • Agent must choose rule from set
         • Act
           • IF true, rules fires and rule is carried out
       • Working Memory
         • Rule can write knowledge to working memory Knowledge may match and fire other rules
       • Various Forms
     • Semantic Networks
       • Syntax
         • Rules to construct sentences
         • Symbols we can use
         • How to combine symbols
       • Semantics
         • How we interpret sentences
         • Assigns meaning to sentences