Conditional Trust Adjustment and Initialization

This paper proposes a rule-based reasoner system that adjusts the ratings based on the underlying siltation, thus, imposes situation awareness to the existing trust models. Furthermore, it helps the trust model to bootstrap by providing an estimation of trust value based on similar situations or similar trustees previously observed. The rules are automatically extracted from the history and encoded as conditions connecting contextual information to trust judgments. Through the use of subjective logic, this method explicitly incorporates uncertainty, thereby making it suitable in situations of partial ignorance and imperfect information. Although our method is suggested for trust and reputation systems, it may also be applied for any other system that uses ratings such as recommender systems. We give an example of implementation of our proposal for a large-scale real dataset.