Topics for Today

Introduction to Knowledge Representation
State Space Search

=====================================

Programming Analogy

Knowledge representations are like abstract data types

Data structure (knowledge structure)
Procedures that operate on the data structure (access, inference, and search procedures)

A representation is a set of conventions about how to describe a class of things (Winston)

=====================================

Parts of a Representation: Syntax and Semantics

Syntax: the symbols that are allowed in the representation's vocabulary, and how they can be combined (notation)

Semantics: the meanings of the objects and relations specified in the syntax

Computational Model: how the objects and relations can be manipulated according to the semantics

Includes: state-space search representations, logical representations, probabilistic representations, etc.

=====================================

Organization: we will begin with state-space search

A search problem: planning a saturday morning

Strategy 1: enumerate all combinations of dept. stores, bookstores, and other places to get Aggie sweatshirts and milk. See which would be best.

Strategy 2: use information about the locations of stores to narrow the search.

Unconstrained search is often not computationally feasible. Intelligent systems (people or machines) use knowledge to make the search problems more tractable.

=====================================