针对LEM2算法存在的局限性,从规则提取的质量和效率两个方面对其进行改进,提出改进LEM2规则提取算法.基于广义决策,定义最简规则优先策略,使得改进后的算法能获取更为简洁的规则,提高规则获取的质量.定义冗余的属性-值对集合,从候选属性-值对集中删除冗余的属性-值对,加快算法的收敛速度,提高规则获取的效率.通过实验证明改进LEM2算法用于不完备信息系统规则提取的有效性. Aiming at the limitations of LEM2 algorithm, this paper improves the quality and efficiency of rule extraction, and proposes an improved LEM2 rule extraction algorithm.Based on generalized decision-making, the most simple rule-first-priority strategy is defined so that the improved algorithm can obtain more Concise rules to improve the quality of rules acquisition.Define the redundant attribute-value pair sets, remove the redundant attribute-value pairs from the candidate attribute-value pair sets, speed up the convergence of the algorithm and improve the efficiency of rule acquisition.According to the experiment It proves that the LEM2 algorithm improves the validity of rule extraction in incomplete information system.