High-Performance Expert System-DBMS Interface for Network Management and Control

Abstract

We discuss the field of network management in modern telecommunications networks along with the need for database and expert systems techniques in this field. We further point out that one of the requirements in such an environment is the capability of inferencing with a very large volume of factual information. In this paper, we present a framework for optimization in coupling an expert system and a database management system (DBMS) in such a way to satisfy this requirement. We adopt a loose coupling strategy as a practical solution allowing utilization of an existing DBMS. Specifically, we present the technique of normalizing the logic query, implementation primitives, and the set of access strategies as mappings between the normalized primitives and implementation primitives. We discuss these techniques in the context of an expert system shell, SQL Inference Engine for Network Applications (SIENA), being developed at IBM Research. We believe that these techniques provide significant insights into the issues in merging expert system and database technologies. © 1989 IEEE