Theory and Algorithm for Rule Base Refinement

摘要

Rule base refinement plays an important role in enhancing the efficacy and efficiency of utilizing a rule base. A rule base concerns three types of redundancies: implication-rule redundancy, abstraction-rule redundancy and dead-end-condition redundancy. This paper proposes two approaches: one is to remove implication redundant rules by using the closure of literal set and the other is to remove abstraction redundant rules by using rule-abstraction. We have developed a software tool to support the first approach. Experiments show that the tool can work correctly and efficiently. The proposed approach can be applied to more application fields.