Prediction of trust relationship based on subjective logic

In social network, trust relationship between two users depends on the trustors opinion about the trustee, and the opinion contains uncertainty, but previous researches did not pay much attention to the uncertainty. Subjective logic is suitable for modeling and analyzing situations involving uncertainty and incomplete knowledge. This paper proposes a method of predicting trust relationship based on subjective logic. Regarding to users typical activities in social network, four types of subjective opinion are defined reflecting different aspects, and non-linear functions are used to map observed values to belief extent. The experimental results on the extended Epinions dataset show that the proposed method can improve the accuracy of trust relationship prediction.