文档介绍：Rqort No. STAN-C’S82-926
Principles of Rule-Based Expert Systems
Department puter Science
Stauford University
Stanford, CA 94305
.
Heuristic Programming Project ’ August 1982
Report No. HPP-82-14
also:
Fairchild
Technical Report
Principles of Rule-Based Expert Systems
BRUCE G. BUCHANAN
HEURISTIC PROGRAMMING PROJECT
DEPARTMENT PUTER SCIENCE
STANFORD UNIVERSITY
RICHARD 0. DUDA
LABORATORY FOR ARTIFICIAL INTELLIGENCE RESEARCH
FAIRCHILD CAMERA AND INS rRUMENT CORPORATION
PALO ALTO, CALIFORNIA
.The Stanford ponent of this research is funded in
part by ARPA contract # MDA903-80-C-0107, NIH contract # NIH RR 0078509,
ONR contract # NO001 4-79-C-0302, and Schlumberger-Doll Research Laboratory.
To appear in M. Yovits ted.) Advances puters, ,
New York: Academic Press
3
i
Table of Contents
1 INTRODUCTION: WHAT IS AN EXPERT SYSTEM? 1
Example: The MYCIN Program 3
ponents 8
2 REPRESENTATION OF KNOWLEDGE 9
Rule-Based Representation Frameworks 11
Production Systems 11
EMYCIN Viewed as a Production System . 13
Alternatives to Rule-Based Representation of Knowledge 15
Frame-Based Representation Languages 16
Logic-Based Representation Languages 16
Generalized Languages 17
Knowledge Representation Issues 17
3 INFERENCE METHODS IN EXPERT SYSTEMS 19
Logical and PlauSible Inference 19
Control 20
Data-Driven Control 20
Goal-Driven Control 21
Mixed Strategies 23
Explicit Representation of Control Knowledge 23
4 REASONING WITH UNCERTAINTY 24
Plausible Inference 24
Bayesian Probability Theory 24
Rules . 26
Uncertain Evidence 27
Certainty Theory 28
Evidence 28
Uncertain Evidence 29
Theory 30
The Dempster/Shafer Theory of Evidence 31
5 KEY CONCEPTS 32
Classes of Problems for Expert Systems 35
The Data 36
The Expertise 37