%0 Journal Article
%A GONG Xiang-yang
%A HU Xiang
%A QUE Xi-rong
%A WANG Bai
%A WANG Wen-dong
%T Social Recommendation Based on Manifold Ranking
%D 2014
%R 10.13190/j.jbupt.2014.03.004
%J Journal of Beijing University of Posts and Telecommunications
%P 18-22
%V 37
%N 3
%X <p>A new recommendation method based on manifold ranking and social matrix factorization is proposed, in which the social similarities among users are calculated by means of manifold ranking, the objective function of ratings matrix factorization is constructed via the regularization technique, with the differences among users' preferences as the penalty of objective function, the social similarities are infused into the low-rank matrix factorization. Experiments show that this method achieves higher precisions and lower root mean square error/mean absolute error (RMSE/MAE) value than other that of cognate methods.</p>
%U https://journal.bupt.edu.cn/EN/10.13190/j.jbupt.2014.03.004