A framework for optimization in expert system-DBMS interface

The authors present a framework for optimization in coupling an expert system and a database management system (DBMS) in such a way to satisfy the requirement of being capable of inferencing with a volume of information that far exceeds main memory capacity. This technique, an extension of traditional relational query optimization, is based on the ideas of normalization and access mapping between the primitives representing the logic query and the implementation primitives. The authors discuss this technique in the context of an expert system shell called SQL inference engine for network applications (SIENA), which is currently being developed