Abstract: To deal with multi-attribute group decision making problems with multi-granularity linguistic phrases of attribute values and attribute weights, a method based on subjective-objective attribute weights integrated and extended Vlsekriterijumska Optimizacija I Kompromisno Resenje （VIKOR） is proposed. The multi-granularity evaluation matrices are unified into the same granularity by the basic linguistic term set, The group evaluate deviations to every attributes are calculated according to the translated linguistic decision matrixes, then the objective weights are obtained by the consistency principle of group evaluation opinions, subjective weights of every attributes are calculated by using the Two-tuple Weighted Averaging （T-WAA） operator, thus, subjective weights and objective weights of attributes are concentrated into comprehensive weights. Evaluation matrixes given by all decision makers are aggregated into group evaluation matrix and its export matrix. Based on extended VIKOR method, the compromise rank result of alternatives is obtained from the group utility values, individual regret values and comprehensive evaluation values. An illustrative example analysis shows the effectiveness and feasibility of this method.

Key words: multiple attributes group decision making, two-tuple linguistic, subjective-objective weight, extended Vlsekriterijumska Optimizacija I Kompromisno Resenje（VIKOR） method

摘要： 针对评价信息、属性权重均为不同粒度语言短语的多属性群决策问题，提出一种基于主客观权重集成及扩展多准则协调优化解（VIKOR）的多属性群决策方法。由基本语言评价集实现对多粒度语言评价矩阵的一致化，基于同一粒度的语言决策矩阵计算群体对属性的评价偏差，基于群体评价意见的一致性原则得到属性客观权重，通过二元语义加权算术平均（T-WAA）算子得到属性主观权重，从而集成主、客观权重求得属性综合权重。集结转化后的单个评价矩阵得到群体评价矩阵及其导出矩阵，由扩展VIKOR方法，根据群效用值、个体遗憾值及综合评价值分别对方案进行排序，获得折衷方案。算例分析表明该方法的有效性与可行性。

关键词: 多属性群决策, 二元语义, 主客观权重, 扩展多准则协调优化解（VIKOR）方法