MCN and MO: Two Heuristic Strategies in Knowledge Compilation Using Extension Rule

EPCCL theory is a target language in knowledge compilation. Knowledge compilation using extension rule (KCER) is a method for knowledge compilation and takes EPCCL theory as target language. Since the size of the compiled knowledge base directly influences the efficiency of on-line queries, in this paper, we present two heuristic strategies aimed at minimizing the size of the compiled knowledge base. They are MCN and MO applied respectively to choose clause and variable. Experimental results show that the two heuristic strategies play a great role in minimizing the size of the compiled knowledge base. The sizes gained by adding MCN and MO to KCER are 3.0-32.9 times less than the sizes gained by KCER without any heuristic.