Harnessing heterogeneous social networks for better recommendations: A grey relational analysis approach

Most of the extant studies in social recommender system are based on explicit relationships, while potential implicit relationships heterogeneous networks remains largely unexplored. This study proposes a new approach to designing by employing grey relational analysis networks. It starts with establishment through user-item bipartite graph, user network graph and user-attribute graph; then uses identify which incorporated into matrix factorization model. Five experiments were conducted test performance our against four state-of-the-art baseline methods. The results show that compared methods, can effectively alleviate sparsity problem, because provides richer information. In addition, method has advantage low requirements for data size efficiently relieves cold start problem. Furthermore, saves processing time, thus increases recommendation efficiency. Overall, proposed improve accuracy rating prediction recommendations provide accurate efficient service users.