Abstract: Aimed at the multi-attribute large group decision-making problem that the decision-makers’ preference information are probabilistic language information in social network analysis, this paper proposes a new probabilistic language multi-attribute large group decision-making method based on group consistency. First, for the situation that decision-makers belong to multiple community clusters to participate in decision-making, this paper proposes the concept of community sub-module membership degree and preprocesses it. Next, considering the social network relationship between decision-maker, it improves the Louvain community detection algorithm to divide the community of large group decision makers. Then, it extends the maximum consensus sequence mining algorithm to the field of probabilistic language information, and proposes a consensus measurement method and consensus achievement model based on group consistency to obtain the final solution. Finally, it applies an example to verify the effectiveness of the proposed method, and illustrates its rationality and advantages by comparative analysis.

Key words: social network analysis；large group decision-making, probabilistic language, overlapping communities, group consistency

摘要： 针对社会网络环境下，决策者的偏好表达为概率语言信息的多属性大群体决策问题，提出一种新的、基于群体一致性的概率语言多属性大群体决策方法。首先，针对现实决策中存在决策者隶属于多个社区集群参与决策的情况，提出社区子模块隶属度的概念并对其进行预处理；其次，通过改进Louvain社区探测算法对群体决策者进行聚类分析；然后，将最大共识序列挖掘算法拓展至概率语言多属性群决策领域，提出一种基于群体一致性的共识测度方法及共识达成模型；最后，通过算例验证本文方法的有效性，并进行对比分析说明本文方法的合理性及优势。

关键词: 社会网络分析, 大群体决策, 概率语言, 重叠社区, 群体一致性