A Novel Trust Evaluation Model Using Graph Clustering Approach

This is the time of rapidly development of electronic relationships between the users of a social network. Any entity in e-interactions has to make decision about trust/distrust to others with respect to the data available on the network. Also the lack of significant information about the entities becomes a challenge that any trust evaluation model has to deals with it. In this paper, a model for evaluation trust with respect to the users’ feedbacks is proposed. The model, is based on a unique generated trusted graph which is the result of applying a proposed initial trust value metric. Also, the communities consisting the network is detected using Markov Clustering Algorithm (MCL). This paper also presents a categorical-based approach for trust evaluation. The proposed model has been compared to another trust metric which is proposed by another paper and. The results, which achieved using Normalized Root Mean Squared Error (NRMSE), show the effects of proposed initial trust value and proposed final trust rate on the final trust evaluations. By more affecting the final trust rate, the model goes more closely to the basis trust metric used for comparing results.