Efficient compilation of large rule bases using logical access paths

Abstract

This paper presents an efficient way to compile and store large deductive databases. Deductive rules are modeled through logical access paths and then integrated into a global structure, the Rule-base Access Path Schema (RAP Schema). The RAP Schema stores information on the interrelationships that exist among the computations required for rule executions. Ground data as well as data derived through some rule executions and cached for later reuse are also modeled in the RAP Schema. Some of the advantages of using this rule base organization scheme include the automatic detection of shared computations, the inclusion of cached results of rule executions and query results in the knowledge base. The details of the compilation and query processing phases are the focus of the paper.