Concept Representation and Trust Relationship Modeling in Fuzzy Social Networks

Social networks (SNs) are changing all aspects of people's way of life, especially their decision making and behavioral styles. Trust, as an essential and important relationship in social network analysis, has gained increasingly more focus. Furthermore, it is important to design an accurate representation and computational model for a trust-enhanced social network. To develop the practical applications of social network analysis, we compare and discuss the properties of trust in SNs and propose the main challenges to measure trust. A fuzzy context-based social network description model is proposed based on these challenges. Multigranularity linguistic variables are used in this model to describe trust relationships among agents. Trust relationship is mapped to a tuple that is named the trust score and contains two parameters: the degree and the strength of trust. We design a trust propagation operator, using t-norm and t-conorm, to estimate the trust propagation score. Then, a trust relationship model for group decision making in the new social network environment is proposed. Finally, an illustrative example of group decision making with incomplete preference information in SNs is given. We show how to use trust relationship to estimate unknown evaluations and complete group decisions in this example. The proposal can realize qualitative descriptions and quantitative measures of trust in social networks. The main differences or innovations of our trust-enhanced social network model are that we distinguish trust relationships according to context and quantify uncertainty in the trust network with the paradigm of computing with words.