Abstract: In the process of logistics supplier selection, to address the problem of incomplete information of expert evaluation under the environment of distributed evaluation language, this paper proposes a group decision-making method under social network considering information complementation. Firstly, a new trust transfer model is proposed to improve the trust value between experts by considering the size of the possibility of experts accepting indirect trust relationships. Secondly, the Jensen-Shannon scatter is firstly extended to the distance metric of the distributed evaluation language, which is used to measure the similarity between experts. Thirdly, based on the K-approximation algorithm, an improved method of completing the residual evaluation information is designed; and then, the expert information is assembled and the consensus metric and feedback adjustment mechanism are constructed to obtain the group decision-making method under the social network. Information is assembled and consensus measure and feedback adjustment mechanism are constructed to obtain the group decision matrix, and the improved EDAS (evaluation based on distance from average solution) method is applied to rank the solutions. Finally, the feasibility and effectiveness of the group decision-making method is verified by the comprehensive evaluation example of logistics service providers.

Key words: indirect trust acceptance possibility, trust transfer model, information completion, distributed language, Jensen-Shannon divergence, group decision-making, consensus

摘要： 在物流供应商选择过程中，针对分布式评价语言环境下专家评价信息不完整问题，提出社会网络下考虑信息补全的群决策方法。考虑专家接受间接信任关系可能性的大小，提出一种新的信任传递模型来完善专家间的信任值；首次拓展Jensen-Shannon散度到分布式评价语言距离度量上，用于衡量专家之间的相似度；基于K-临近算法，设计改进的残缺评价信息补全方法；对专家信息进行集结并构建共识度量与反馈调节机制，得到群决策矩阵，并运用改进的EDAS（evaluation based on distance from average solution，离平均方案（平均解）距离）方法对方案进行排序；通过物流服务供应商综合评估算例验证该群决策方法的可行性和有效性。

关键词: 间接信任接受可能性, 信任传递模型, 信息补全, 分布式语言, Jensen-Shannon散度, 群体决策, 共识