Trust evaluation using items ratings in online communities

With the increasing of information overloading, it is necessary to have a robust trust model to help the users to collect the reliable information. Recent researches in trust prediction are extremely rely on users explicit trust, which is based on users past experiences. However, users explicit trust is not always available and if it is available, it will be so sparse and can't be used to predict the trust between two unknown users with high accuracy. In this paper, we propose an approach to predict trust values between users based on items ratings and without using explicit trust values. Using this model, we can predict trust values without access to web of trust and then apply it in online communities. We have investigated our approach with some experiments with real-world dataset collected from epinions.com. The results show that our approach can predict trust values with high accuracy.