Subjective implicit trust evaluation in online communities

A robust trust model helps the users to collect reliable information and overcome the information overloading. Recent researches in trust prediction are extremely rely on users explicit trust, which is based on users past experiences. However, users´ explicit trust ratings are not always available and when available, it is very sparse and cannot be used to predict the trust between two unfamiliar users with high accuracy. Therefore, the need for an accurate implicit trust model to calculate trust values between users without any explicit trust network is undeniable. Furthermore, trust is a subjective concept and trust ratings given to a user by others may vary according to the mentality of the raters even when they have similar experiences with a user. In this paper, we propose a subjective implicit trust model to predict trust values between users based on their ratings to different items and without using explicit trust values. Moreover, we compare our approach with another model that did not consider the subjectivity of trust and we show the importance of taking into account the subjectivity in calculating the trust values. We have investigated the accuracy of our approach with some experiments with a real-world dataset collected by epinions.com. The results show that our approach can calculates trust values with high accuracy.