Abstract

Rules are widely used in present-day intelligent systems for modeling intelligent behavior, building expert systems, database rule systems, and advanced document retrieval systems. Most rule-based system programs are extremely computation-intensive and run quite slowly. This problem may get still worse when a rulebase does not fit into primary memory completely. This article reviews currently existing rule matching and indexing techniques. This report is organized into two parts. Part I presents the well known rule matching algorithms, Rete and TREAT, along with their performance evaluations. Part II presents rule indexing techniques for large scale rulebases. Here two sources of indexing are presented- attribute level and value level indexing- which improve the run time response of the rule-based systems. A forward chaining inferencing using the attribute level indexing is also presented.

Department(s)

Computer Science

Comments

The Author is a Graduate Student.

Keywords and Phrases

Artificial Intelligence, Database Rule Systems, Rule Indexing, Rule Clustering, Search Strategies, Rule-base.

Report Number

CSc-95-09

Document Type

Technical Report

Document Version

Final Version

File Type

text

Language(s)

English

Rights

© 1995 University of Missouri - Rolla, All rights reserved

Publication Date

1 July, 1995

Download

Share

 
COinS