Abstract: Aiming to improve the decision-making precision of Analytic Hierarchy Process（AHP），an improved comprehensive method is proposed by means of the knowledge of information entropy and mutual information in rough sets theory.This method begins with constructing an objective judgement matrix by attribute significance defined by information entropy，and then gets a combinatorial judgement matrix with the objective judgment matrix and the subjective judgment matrix derived from AHP method，therefore establishes an optimization model which can determine the combinatorial judgement matrix.Finally，decision-making is guided by the combinatorial matrix’s weight value determined by the optimization model.An illustrative example is given to show the higher decision precision，feasibility and effectiveness of this comprehensive method.

摘要： 借助于粗糙集理论中知识信息熵与互信息的知识，提出了一种用于改进层次分析法决策精度的改进的综合决策方法，该方法首先通过信息熵定义的属性重要度构造出客观判断矩阵，在此基础上将该客观判断矩阵和层次分析法中的主观判断矩阵相结合以得出一组合判断矩阵，并为此建立了一个能确定组合判断矩阵的最优化模型，最后通过由该最优化模型确定出的组合判断矩阵的权重来指导决策。算例表明该方法是可行的，能提高决策的准确度。