A clustering and fusion method for large group decision making with double information and heterogeneous experts

In real large group decision-making (LGDM) problems, there usually exist double information consisting of decision basis information and preference relation information, and heterogeneous experts who have different and uncertain opinions about criteria importance. Aiming at this situation, we propose an expert clustering and information fusion method. First, criteria weights of experts are derived through an optimization model based on the minimization of the deviation between double information. Then, a double clustering method combining the similarity degrees of experts' fuzzy preference relations and of their criteria weights is proposed to classify the large-scale experts into several clusters. Meanwhile, a clustering validity index is put forward to objectively determine the clustering threshold to ensure its rationality. Third, targeting to guarantee high consistency and consensus of the final selected alternative, an optimization model is constructed to fuse the double information of clusters to obtain the collective comprehensive information. Finally, a case study and comparisons are performed to verify the feasibility and effectiveness of the proposed method. The comparison results suggest that the heterogeneity of experts about criteria weights is important in LGDM, the clustering results are more rational when combining the similarities of the fuzzy preference relations and the criteria weights, and the consistency and consensus indexes of the fusion result are greater with the proposed optimization model.

Automated summary

This study proposes an expert clustering and information fusion method to address the issue of double information and expert heterogeneity in large group decision-making. By deriving criteria weights through an optimization model, clustering experts based on similarity degrees, and constructing a fusion model for collective information, the method aims to ensure consistency and consensus in the final alternative selection process.