Abstract: An approach based on two-tuple is put forward to solve the hybrid multiple attribute group decision making problem with weight information unknown. In order to solve the weights of experts, an approach based on educed vector is proposed to do clustering analysis. Then combine both weight value inside of the group and weight value between the groups. Based on this, aggregate the experts’ imformation into two-tuple linguistic matrix. According to maximizing deviation, the attribute weights can be determined, then the value of projects can be obtained. Compare the value of every project and order them. Finally, an example illustrates of missiles, using the proposed method to solve hybrid multiple attribute group decision making problems, demonstrate the procedure of the method and show this method is easy to operate and can provide abundant decision-making information for decision makers.

Key words: hybrid multi-attribute, group decision making, clustering analysis, two-tuple, maximum deviation

摘要： 为了解决专家权重、指标权重均未知的混合型多属性群决策问题，提出基于二元语义的多属性群决策方法。首先，针对专家权重问题，提出了运用判断矩阵导出向量进行聚类分析，并综合类间、类内权重得出结果。在此基础上，集结多名专家意见得到每个方案的二元语义矩阵，根据离差最大化原理，客观确定属性的指标权重，求出每个方案的值。根据数值大小比较排序；最后以多种型号导弹作为算例，用所提出方法对混合型多属性决策问题进行求解，结果表明该方法易于操作且可为决策者提供丰富的决策信息。

关键词: 混合多属性, 群决策, 聚类分析, 二元语义, 离差最大化