Social recommender system by embedding social regularization

The rapid growth in information on the World Wide Web has created different challenges to the users such as finding relevant and useful information and knowledge. To deal with such problem, recommender systems came into existence. Collaborative filtering techniques have gained much more popularity than other techniques in recommender system. As in real life, we ask our friends for different recommendations. But traditional systems ignore social relationships among users. In order to resolve this problem and improve recommender system´s results, the idea of using social recommender system was discussed which contains the capability of identifying user´s interests and preferences and their social network relationship. However, this approach is not sensitive to those users whose friends have dissimilar tastes. To tackle the problem of inaccuracy in result due to information deficiency, the social regularization term is used to impose constraints between one user and their friends individually.