IEEE Trans Cybern. 2023 Jun;53(6):3748-3759. doi: 10.1109/TCYB.2022.3170589. Epub 2023 May 17.
Consensus reaching process (CRP) is a key topic in the area of group decision making (GDM). When the consensus level is not high enough, it becomes necessary to adjust the original opinions of decision makers (DMs). To offer the adjustment reference for DMs, we build the programming models to determine the minimum modification to be carried out from the individual and global perspectives. Meanwhile, all DMs are divided into two subgroups: DMs with acceptable and unacceptable consensus levels. If some DMs with unacceptable consensus level do not accept the relevant modifications, the Nash bargaining game-based programming model is built for the fairness and efficiency of modifications. When some DMs refuse to make any modifications or tend to modify the opinions in their way, with respect to different group consensus situations, we make the minimum hybrid penalty mechanism by the Nash bargaining game-based programming models. For each case, we determine the corresponding optimal modification mechanism in view of the fixed individual total modification and the maximum consensus level. Furthermore, we study the arrangements of weights of DMs according to their cardinal and ordinal consensus contributions. Based on these results, we present a new algorithm and illustrate its application by a numerical example. Moreover, we carry out the sensitivity and comparison analysis. We summarize the conclusions and future research directions in the end. The main originality of the new method includes: the fairness and efficiency of modifications, and the determination of the hybrid penalty mechanism.
达成共识的过程(CRP)是群体决策(GDM)领域的一个关键主题。当共识水平不够高时,就需要调整决策者(DM)的原始意见。为了为 DM 提供调整参考,我们从个体和全局角度构建了确定要进行的最小修改的编程模型。同时,将所有 DM 分为两组:具有可接受和不可接受共识水平的 DM。如果一些具有不可接受共识水平的 DM 不接受相关修改,则为公平性和修改效率构建基于纳什讨价还价博弈的编程模型。当一些 DM 拒绝进行任何修改或倾向于以自己的方式修改意见时,针对不同的群体共识情况,我们通过基于纳什讨价还价博弈的编程模型制定最小混合惩罚机制。对于每种情况,我们根据固定的个体总修改量和最大共识水平确定相应的最优修改机制。此外,我们根据 DM 的基数和序数共识贡献来研究 DM 的权重安排。基于这些结果,我们提出了一种新的算法,并通过数值示例说明了其应用。此外,我们进行了敏感性和比较分析。最后,我们总结了结论和未来的研究方向。该新方法的主要创新点包括:修改的公平性和效率,以及混合惩罚机制的确定。