Nowadays, the increasing complexity of the social environment brings much difficulty in group decision making. The more uncertainty exists in a decision-making problem, the more collective wisdom is needed. Therefore, large scale group decision making has attracted a lot of researchers to investigate. Since the probabilistic linguistic terms have impressive performance in expressing DMs’ opinions, this paper proposes a novel method for large scale group decision making with probabilistic linguistic preference relations. More specifically, (1) a probability k-means clustering algorithm is introduced to segment DMs with similar features into different sub-groups; (2) an integration method is proposed to construct the collective probabilistic preference relation that retains initial information to the most extent; (3) taking the personality of each DM into account, a consensus model is constructed to improve the rationality and efficiency of consensus reaching process. Several simulation experiments are designed to analyze the influence factor in the feedback mechanism and make some comparative analysis with the existing method. Finally, an illustrative example of contractor selection is conducted to verify the validity of the proposed method.

如今，日益复杂的社会环境给团队决策带来了很多困难。决策问题中存在的不确定性越多，就需要越多的集体智慧。因此，大规模的群体决策吸引了许多研究者进行研究。由于概率语言术语在表达决策者的意见方面具有令人印象深刻的表现，因此本文提出了一种新的具有概率语言偏好关系的大规模群体决策方法。更具体地说，（1）引入了概率k均值聚类算法，将具有相似特征的DM划分为不同的子组；（2）提出了一种整合方法，以建立最大程度地保留初始信息的集体概率偏好关系；（3）考虑到每个决策者的个性，建立共识模型以提高共识达成过程的合理性和效率。设计了几种仿真实验来分析反馈机制中的影响因素，并与现有方法进行比较分析。最后，以承包商选择为例，验证了所提方法的有效性。