Recommending a method based on hybrid individual and group trust in collaborative filtering

The huge volume of information on the Internet, generates a lot of problems for users of commercial websites and social networks; so that the competition between their owners in providing appropriate, sufficient and reliable information for their users has been stimulated. Recommender systems using collaborative filtering, have been emerged to guide and support users in choosing the needed product among the massive volume of information. Recommender systems based on trust, using the grants of trust or combining them with the grants of similarity, provide some recommendations. In a situation where user receives the suggestions from people whom he knows or are similar to himself, has more trust comparing with the situation where the suggestion is just based on similarity with others. In this paper, a hybrid recommender system, based on trust with individual and group views is presented. Individual trust is calculated based on item points which is earned by users. Group trust also, with regard to a particular method in clustering, puts in one group the users who have similar tastes in selecting items. Finally, the system puts these two views beside the factor of similarity and proposes to offer. Assessments carried out in this paper show that the proposed model compared to other models of collaborative filtration, has significantly increased the prediction accuracy.