推理规则链的确定性构建

Abstract

推理是人工智能领域研究的重点。推理过程一般是非确定的，会产生许多冗余的推理分支。本文针对专家系统中常用的产生式规则，提出了一种确定性构造推理链的方法。方法运用逻辑代数理论，首先确立了规则集与逻辑函数的对应关系，证明推理规则链中的所有规则对应构成逻辑函数质蕴含集中的一个特定最小子集，给出了确定该子集的基本算法。子集中质蕴含对应的规则按逻辑顺序排列即构成了一条合理的推理规则链。 Reasoning is a research focus of artificial intelligence. Uncertainty in reasoning process generally produces many redundant reasoning branches. Based on production rule commonly used in expert system, this paper proposes a method to build the reasoning chain determinately. The method uses the theory of logic algebra. First we establish correspondence between a set of production rules and a logical function, and then prove that rules in a reasoning chain corresponding to a special minimal subset of the prime implication set that consists of the logical function, finally give out the basic algorithm to determine the subset. A sound reasoning chain is formed by arranging rules corresponding to prime implications in subset according to the logical order.