A Speedup Algorithm for Repetition of Hypothetical Reasoning

This paper presents performance evaluations of a speedup method for hypothetical reasoning to a firstorder predicate logic knowledge base in the case when reasoning is executed repeatedly replacing hypothetical knowledge and keeping background knowledge and the goal intact. The proposed method consists of substituting hypotheses for predicate knowledge which has recursive structures, deriving bit patterns of solutions from background knowledge and the goal in advance when all the hypotheses are true, and finding actual solutions from inclusion relation with bit patterns of hypothetical knowledge.