An uncertainty-aware computational trust model considering the co-existence of trust and distrust in social networks

Abstract

There are few trust models capable of incorporating the co-existence of trust and distrust as distinct concepts. In this regard, most of the existing trust models implicitly use distrust parameters to refine and calculate trust values. However, recent studies have indicated that trust and distrust are two distinct but co-existing concepts. In other words, although trust and distrust are constructed based on different characteristics, they can be used together in the decision making and recommendation processes. In this paper, we present a trust-distrust model for social networks considering subjective and objective characteristics of trust and distrust simultaneously. Competence, honesty, satisfaction, similarity, motivation, availability, tendency to be trusted, the existence of long-term connection/friendship, and centrality are the trustworthiness characteristics covered by the model. Also, surprisal, dishonesty, dissatisfaction, conflict degree, account lifetime, and sudden changes in the number of friends, likes, and comments are the distrust characteristics considered by the model. The proposed model takes into account the uncertainty, sharpness, and vagueness of the beliefs by using subjective logic. The results of the conducted evaluations demonstrate that the proposed model is highly accurate in the decision-making process and has a 90% accuracy in calculating the trust and distrust. We have also compared the results with other similar approaches, by which the proposed model showed a 34% improvement.