Managing minority opinions in large-scale group decision making based on community detection and group polarization

As the decision environment becomes increasingly complicated, the demand for large-scale group decision making (LSGDM) is increasing. Because of the differences in decision-makers' (DMs') personalities, knowledge, and experience, and incomplete information, irrational decision making behaviors, and minority opinion situation frequently appears. This study considers this phenomenon and proposes an LSGDM method by developing a similarity network of DMs on the basis of incomplete preference information. Moreover, the large number of DMs is divided into several clusters using a community detection model, and minority opinions are identified. To show the attitudes of DMs in different decision scenarios, the group polarization effects of individuals within communities are analyzed, and minority opinions using the family aggregation operators of the ordered weighted averaging operator are managed. Given the massive user involvement in online restaurant ratings, the recommendation list of Dianping.com is determined using the proposed LSGDM method. The results of the sensitivity and comparative analyses show that the LSGDM method is flexible and applicable because the attitude of DMs is considered.