PATTERN MATCH REDUCTION IN A KBMS VIA THE MAINTENANCE OF DERIVED INTERMEDIATE RELATIONS

A pattern match reduction technique is presented for use in a system w hich integrates relational DBMS technology and AI production system te chnology. Production systems in general provide only rudimentary data storage and access capabilities. The integration of a DBMS with a prod uction system would provide the production system access to the sophis ticated data storage and retrieval capabilities of a DBMS and, at the same time, allow the application of production system inferencing capa bilities to existing conventional databases. In an integrated system t he left hand side of production rules can be represented as a relation al query. The technique presented here maintains state information acr oss production system cycles by storing and updating derived relations representing partial matches at internal nodes in the query tree.