Using Integrity Constraints to Control Search in Knowledge Base Systems
Abstract
A knowledge base system combines data management and reasoning capabilities. It can be used to store and manipulate a body of knowledge which consists of a set of rules and facts. The rules and facts in a knowledge base capture syntactic information. In contrast, integrity constraints contribute semantic information about the represented knowledge. They impose restrictions on the states of the world that a knowledge base can model. Typically, integrity constraints are used to update and maintain the knowledge base, but they can also be a powerful tool for answering queries. A logic programming system augmented with constraint processing, data storage, and data manipulation capabilities forms the basis for a knowledge base system. Both a runtime approach and a compiled approach to using integrity constraints in logic programming systems to identify and eliminate unproductive search activity have been implemented within an existing parallel logic programming system, PRISM. We have thus extended PRISM to be a testbed for knowledge base applications. The extended system provides the basis for a series of experiments which 1) compare the performance of the compiled approach and the runtime approach, 2) demonstrate that significant classes of knowledge representation domains can use integrity constraints effectively, and 3) reveal additional techniques necessary to put the theoretical approaches into practice. We show that using constraints to process a query can reduce search space and response time. Furthermore we show that the compiled approach reduces response time much more than the runtime approach.
Document Details
- Document Type
- Technical Report
- Publication Date
- Feb 01, 1990
- Accession Number
- ADA221291
Entities
People
- Anne Litcher
- Jack Minker
- Mark Giuliano
- Theresa Gaasterland
- Yuan Liu
Organizations
- University of Maryland