روجیار پیرمحمدیانی

Although trust relations among users in social media are the evidence of social influence, but the level of this impact depends on strength of trust among users. Therefore, in this paper we study the effect of trust relationships’ strength among users through the development of matrix factorization method. Matrix factorization is a method to predict the rate that user assign to the products. For this purpose, based on similarity factors the strength of trust relationship is measured, then this feature as an effective parameter was added to matrix factorization. The proposed model is applied on Epinions data set. The weight of trust relationships’ strength parameter is determined through the proposed model. Based on the results, by adding the strength of trust relationship the error rate is reduced. Furthermore, if the strength of trust between users is larger than 0.8, the error reduction will be more significant. Improving the results through more accurate estimation shows that this effect is set correctly.