Topic-Aware Experience Trust Computation With Refined Interaction In Social Networks

Abstract

Analysing data in social networks to discover trustworthiness among users has become an important topic in social computing in the recent years. Trust is a term which is used to measure reliability among peers or users during their interaction process in distributed systems. It has been widely studied and applied in various systems such as searching, multiagent, decision support, virtual asistant or recommender systems. However, majority of the current studies rely only on the network structure, some form of communication but neglect the various types of interaction and the contexts of interaction such as user intesrests. In this paper, wepropose a computational model of topic-aware experience trust on social networks, which is a function of user interest degrees and refinement of interaction types including degrees of familiarity, dispatches and responds. We perform experiments to investigate how factors of interaction and the proposed interest measures impact trustworthiness. Our experimental results show that degrees of user interest in topics affect trust estimation more than interaction and the proposed refined experience trust significantly outperforms the trust estimation with the single form of interaction.