A Granular Computing-Driven Best–Worst Method for Supporting Group Decision Making

Abstract

In group decision making (GDM), there are seldom ideal scenarios that all the preference information given by all individuals reach a highly level of agreement. Conflicts are present in the information fusion process and decision makers (DMs) have to negotiate and reconcile differences. To address this issue, it becomes inevitable to consider intelligent GDM method. In this article, we propose the granular neural network (GNN) to realize the aggregation process from the perspective of granular computing and machine learning. Our study is involved in an extension of best–worst method to the GDM scenario. The procedure is outlined as follows: first, information granules are allocated around the prototype of individuals’ preferences, complying with the principle of justifiable granularity. Thereby, the granular inputs are brought into a well-trained GNN. An adaptive particle swarm optimization algorithm is applied to optimize allocation of information granules. We calculate the threshold of consistency index for this granular model. Finally, a case study about hotel selection on Booking.com is presented to illustrate the performance of the proposed model. In addition, we use the stochastic analysis method to randomize the weights of group members with the objective to assess the robustness of the model. The feasibility and validity of the model are demonstrated by completing comparative analysis. The originality of this article is to establish a real data-driven granular GDM model both considering the optimization of group consistency and consensus.