A Novel Linguistic Cohesion Measure for Weighting Experts’ Subgroups in Large-Scale Group Decision Making Methods

摘要

Today, decision making problems are continually evolving due to the new needs of society, caused mainly by continuous technological advances. Many times, in order to face these new decision problems, it is no longer enough with the participation of only a few experts, but hundreds or thousands are necessary. The engagement of many experts implies, in turn, the appearance of new challenges such as the management of greater uncertainty, scalability, opinions’ polarization etc. This contribution is focused on group decision making problems in a large-scale context in which uncertainty is modeled by linguistic information and how to deal with the scalability problem under these conditions through clustering methods. Clustering methods are used to manage the scalability problem by grouping the initial large group of experts into smaller subgroups according to the similarity of their opinions and assign different weights to such subgroups. The weights assignment is a key issue due to its influence in the final solution of the problem and classically, it has been carried out by taking into account exclusively the size of the subgroups, by ignoring other features such as the cohesion of the opinions in the subgroups, which may provoke a misassignment of the importance in experts’ subgroups and thus, unfair solutions. Therefore, this paper introduces a new way to calculate and assign properly the relevance of experts’ subgroups in clustering methods by taking into account both size and cohesion of such subgroup under uncertainty conditions in which experts use linguistic assessments.