针对粗糙集理论核心内容之一的知识约简问题,该文提出了一种改进的互信息的属性约简算法。该算法结合信息论中信息熵与互信息的概念定义了粗糙集里的一种新的属性重要度,并以此属性重要度为启发式信息进行属性约简。实验分析表明:在大多数情况下,该种算法都能够得到决策表的最小约简。 Aiming at the problem of knowledge reduction of one of the core contents of rough set theory, this paper proposes an improved attribute reduction algorithm of mutual information. The algorithm combines the concept of information entropy and mutual information in information theory to define a new attribute importance degree in rough set, and uses this attribute importance degree to heuristic information to attribute reduction. Experimental analysis shows that in most cases, the algorithm can get the minimum reduction of the decision table.