摘要
针对BPR模型收敛速度慢的问题,Randle S提出一种非均匀采样非隐式反馈数据方法 AOBPR模型来加快收敛速度,可是该算法只能利用隐式反馈数据.为了改进其算法的不足,我们提出了一种将AOBPR模型与经典的基于矩阵分解的SVD++算法相结合的算法AOBPR_SVD++.改进后的算法不仅能利用隐式反馈数据也能利用显式反馈数据.最后通过在两个真实数据集中进行实验验证,表明改进后的算法可以获得更好的推荐效果.
In order to solve the problem of slow convergence rate of BPR models,Randle S proposed a method of non-uniform sampling non-implicit feedback data named AOBPR.However,the algorithm could only use implicit feedback data.In order to improve the algorithm,we proposed a new algorithm combining the AOBPR model with classical SVD++ algorithm based on matrix factorization.The improved algorithm could not only use implicit feedback data but also explicit feedback data.The experimental results in two real data sets show that the improved algorithm can obtain better recommendation results.
引文
[1]朱扬勇,孙婧.推荐系统研究进展[J].计算机科学与探索,2015,9(5):513-525.ZHU Y Y,SUN J.Recommender system:Up to now[J].Journal of Frontiers of Computer Science and Technology,2015,9(5):513-525.
[2]黄震华,张佳雯,田春岐,等.基于排序学习的推荐算法研究综述[J].软件学报,2016,27(3):691-713.HUANG Z H,ZHANG J W,TIAN C Q.Survey on learning-to-rank based recommendation algorithms[J].Journal of Software,2016,27(3):691-713.
[3]李改.融合显/隐式反馈的协同排序算法[J].计算计应用,2015,35(5):1328-1332,1341.LI G.Collaborative ranking algorithm by explicit and implicit feedback fusion[J].Journal of Computer,2015,35(5):1328-1332,1341.
[4]RENDLE S,FREUDENTHALER C,GANTNER Z,et al.BPR:Bayesian personalized ranking from implicit feedback[C]//Proceedings of the 25th Conference on Uncertainty in Artificial Intelligence(UAI09).Arlington,Virginia,USA:AUAI Press,2009:452-461.
[5]PAN W,LI C.GBPR:group preference based Bayesian per-sonalized ranking for one-class collaborative filtering[C]//Proceedings of the 26th International Conference on Knowledge Discovery and Data Mining.New York:ACM,2009:667-676.
[6]SHI Y,KARATZOGLOU A,BALTRUNAS L,et al.CLiMF:Collaborative Less-is-More Filtering[C]//Proceedings of the 23rd International Conference on Artificial Intelligence.New York:ACM,2013:3077-3081.
[7]SHI Y,KARATZOGLOU A,BALTRUNAS L,et al.xCLiMF:Optimizing expected reciprocal rank for data with multiple levels of rele vance[C]//Proceedings of the 6th ACM Conference on Recommender Systems.New York:ACM,2013:431-434.
[8]RENDLE S,FREUDENTHALER C.Improving pairwise learning for item recommendation from implicit feedback[C]//Web Search and Data Mining.WSDM:ACM,2014:273-282.
[9]KOREN Y,BEL R,VOLINSKY C.Matrix factorization techniques for recommender systems[J].Computer,2009,42(8):30-37.
[10]KOREN Y.Factorization meets the neighborhood:a multifaceted collaborative filtering model[C]//Proceedings of the 14th ACM SIGKDD International Conference on knowledge Disc overy and Data Mining.New York:ACM,2008:426-434.