In collaborative filtering algorithm, the classical matrix factorization may destroy the adjacent structures among data points from high dimension to low dimension. A novel collaborative filtering algorithm based on structured projective nonnegative matrix factorization (CF-SPNMF) was proposed in order to overcome the problem. The algorithm contains both offline learning and online searching. In the offline learning stage, projective nonnegative matrix factorization was applied to obtain the low dimensional latent factors of user preference without changing the intrinsic structure of users cluster. In the online searching stage, cosine similarity was used to measure the similarity between the target user and training users based on the latent factors inferred in the offline stage. Then the most similar neighbor set was further found. The extensive experiments on real-world data set demonstrate that the proposed CF-SPNMF achieves better rating prediction performance than traditional methods using either matrix factorization or users clustering in original rating matrix．
JU Bin, QIAN Yun-tao, YE Min-chao. Collaborative filtering algorithm based on structured projective nonnegative matrix factorization. JOURNAL OF ZHEJIANG UNIVERSITY (ENGINEERING SCIENCE), 2015, 49(7): 1319-1325.
［1］LU L, MEDO M, YEUNG C H, et al. Recommender systems ［J］. Physics Reports, 2012, 519(1): 149．
［2］SU X, KHOSHGOFTAAR T M. A survey of collaborative filtering techniques ［J］. Advances in Artificial Intelligence, 2009, 2009(10): 424．
［3］LINDEN G, SMITH B, YORK J. Amazon. com recommendations: item-to-item collaborative filtering ［J］. Internet Computing, IEEE, 2003, 7(1): 76-80．
［4］ADOMAVICIUS G, TUZHILIN A. Toward the next generation of recommender systems: a survey of the state-of-the-art and possible extensions ［J］. IEEE Transactions on Knowledge and Data Engineering, 2005, 17(6): 734-749．
［5］KOREN Y, BELL R, VOLINSKY C. Matrix factorization techniques for recommender systems ［J］. Computer, 2009, 42(8): 30-37．
［6］PATEREK A. Improving regularized singular value decomposition for collaborative filtering［C］∥Proceedings of KDD Cup and Workshop. California: ACM, 2007: 58．
［7］ZHANG S, WANG W, FORD J, et al. Using singular value decomposition approximation for collaborative filtering ［C］∥ 7th IEEE International Conference on E-Commerce Technology. Ankara: IEEE, 2005: 257-264．
［8］SREBRO N, JAAKKOLA T. Weighted low-rank approximations ［C］∥ICML. Washington: DBML, 2003: 720-727．
［9］MNIH A, SALAKHUTDINOV R. Probabilistic matrix factorization ［C］∥Advances in Neural Information Processing Systems. Vancouver: MIT, 2007: 1257-1264．
［10］LEE D D, SEUNG H S. Learning the parts of objects by non-negative matrix factorization ［J］. Nature, 1999, 401(6755): 788-791．
［11］HOYER P O. Non-negative sparse coding ［C］∥ Proceedings of the 2002 12th IEEE Workshop on Neural Networks for Signal Processing. London: IEEE, 2002: 557-565．
［12］胡俐蕊,吴建国,汪磊. 线性投影非负矩阵分解方法及应用［J］. 计算机科学, 2013, 40(10): 269-273．
HU Li-rui, WU Jian-guo,WANG Lei.Application and method for linear projective non-negative matrix factorization \[J\].Computer Science,2013,40(10):269-273.
［13］YUAN Z, YANG Z, OJA E. Projective nonnegative matrix factorization: sparseness, orthogonality, and clustering ［J］. Neural Processing Letters, 2009, 2009(1): 33-47．
［14］YANG J, YANG S, FU Y, et al. Non-negative graph embedding ［C］∥ IEEE Conference on Computer Vision and Pattern Recognition. Hongkong: IEEE, 2008: 18．
［15］LIU X, YAN S, JIN H. Projective nonnegative graph embedding ［J］. IEEE Transactions on Image Processing, 2010, 19(5): 1126-1137．
［16］BELKIN M, NIYOGI P. Laplacian eigenmaps and spectral techniques for embedding and clustering ［C］∥NIPS. Vancouver: MIT, 2001: 585-591．
［17］NIYOGI X. Locality preserving projections ［C］∥Neural Information Processing Systems. Los Angeles: MIT, 2004: 153-161．
［18］CAI D, HE X, HAN J, et al. Graph regularized nonnegative matrix factorization for data representation ［J］. IEEE Transactions on Pattern Analysis and Machine Intelligence, 2011, 33(8): 1548-1560．
［19］SEUNG D, LEE L. Algorithms for non-negative matrix factorization ［J］. Advances in Neural Information Processing Systems, 2001, 2001(3): 556-562．
［20］ZHANG Z, ZHAO K, ZHA H. Inducible regularization for low-rank matrix factorizations for collaborative filtering ［J］. Neurocomputing, 2012, 97(1): 52-62．
［21］HU Li-rui, WU Jian-guo, WANG Lei. Application and method for linear projective non-negative matrix factorization ［J］. Computer Science, 2013, 40(10): 269-273.
