SWTrust: Generating Trusted Graph for Trust Evaluation in Online Social Networks

Abstract

To solve the problem of "Can Alice trust Bob on a service in a social network setting", we propose the SWTrust framework to generate trusted graphs for trust evaluation in online social networks: (1) to develop a novel PSN algorithm for preprocessing a social network, where we classify a user's neighbors by their social distance and define neighbors' priority based on their topic-related degree and target-related degree, (2) to develop distributed algorithms to build trust network (BTN) and to generate trusted graph (GTG). Experimental results with data from Epinions.com show that our work can discover short paths and generate high quality trusted graphs efficiently.