近年来，推荐系统在网络平台上得到了广泛的应用，它可以从巨量的数据中提取有用的信息，并根据用户的喜好向用户推荐合适的项目。基于此，提出一种利用相似用户对不同项目的评分数据作为源域对目标域的跨领域推荐表现方法，在研究的目标域中引入项目集群效应，提取与某个项目相关的具有相似特征的信息。该方法可有效解决数据稀疏性的问题，由于目标域的稀疏性，目标域的测试集中多数项目是拥有很少评分的，它们的信息难以从训练集中获得。所提模型的一个优点是，能够通过基于每个用户对项目的评分和缺失情况相关的变量的聚类，将来自缺失机制和特定项目集群特征的信息结合起来。MovieLens数据分析表明，与现有推荐方法和跨领域推荐方法相比，所提出的引入集群效应跨领域推荐新方法在预测精度上有着有效的提升。
In recent years, the recommender system has been widely used in online platforms, which can extract useful information from giant volumes of data and recommend suitable items to the user according to user preferences. In this article, we put forward a crossdomain recommendation method based on the rating data of the different projects from similar users, introducing project cluster effect in the target domain of the study, the use of this specific group of singular value decomposition with the method of extracting information associated with a project with similar characteristics. This method could effectively solve the problem of data sparsity. Due to the sparsity of the target domain, most items in the test set of the target domain have few scores, and their information is challenging to obtain from the training set. A strictly related problem is the one of collaborative filtering in recommender systems, where an algorithm tries to extrapolate missing information about the items from the rating activity of the users in order to provide a specific ad-hoc ranking for each user also on the items that have not been rated (on this see discuss how to aggregate the information from multilayer networks, while showing the importance of centrality measures for this issue). MovieLens data analysis indicated that, compared with the existing recommendation methods and cross-domain recommendation methods, the proposed new method of cross-domain recommendation with cluster effect has a significant improvement in the prediction accuracy.
[1] Mazumder R, Hastie T, Tibshirani R. Spectral regularization algorithms for learning large incomplete matrices[J]. The Journal of Machine Learning Research, 2010, 11: 2287-2322.
[2] Bell R M, Koren Y. Scalable collaborative filtering with jointly derived neighborhood interpolation weights[C]//Seventh IEEE International Conference on Data Mining (ICDM 2007). October 28-31, 2007, Omaha, NE, USA. IEEE, 2008: 43-52. DOI: 10.1109/ICDM.2007.90.
[3] Salakhutdinov R, Mnih A, Hinton G. Restricted Boltzmann machines for collaborative filtering[C]//Proceedings of the 24th international conference on Machine learning. June 20-24, 2007, Corvalis, Oregon, USA. New York: ACM, 2007: 791-798. DOI: 10.1145/1273496.1273596.
[4] Blanco-Fernandez Y, Pazos-arias J J, Gil-Solla A, et al. Providing entertainment by content-based filtering and semantic reasoning in intelligent recommender systems[J]. IEEE Transactions on Consumer Electronics, 2008, 54(2): 727-735. DOI: 10.1109/TCE.2008.4560154.
[5] Lops P, de Gemmis M, Semeraro G. Content-based recommender systems: State of the art and trends[M]// Recommender Systems Handbook. Boston, MA: Springer, 2011: 73-105.10.1007/978-0-387-85820-3_3.
[6] 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. DOI: 10.1109/TKDE.2005.99.
[7] Park S T, Pennock D, Madani O, et al. Naïve filterbots for robust cold-start recommendations[C]//Proceedings of the 12th ACM SIGKDD international conference on Knowledge discovery and data mining. August 20-23, 2006, Philadelphia, PA, USA. New York: ACM, 2006: 699-705. DOI: 10.1145/1150402.1150490.
[8] Zhu Y Z, Shen X T, Ye C Q. Personalized prediction and sparsity pursuit in latent factor models[J]. Journal of the American Statistical Association, 2016, 111(513): 241-252. DOI: 10.1080/01621459.2014.999158.
[9] Zhao L L, Pan S J, Xiang E W, et al. Active transfer learning for cross-system recommendation[C]//Proceedings of the Twenty-Seventh AAAI Conference on Artificial Intelligence. July 14-18, 2013, Bellevue, Washington. New York: ACM, 2013: 1205-1211. DOI: 10.5555/2891460.2891628.
[10] Li B, Yang Q, Xue X Y. Can movies and books collaborate?: cross-domain collaborative filtering for sparsity reduction[C]//Proceedings of the 21st International Joint Conference on Artificial Intelligence. July 11-17, 2009, Pasadena, California,.
[11] Ren S T, Gao S, Liao J X, et al. Improving cross-domain recommendation through probabilistic cluster-level latent factor model[C]//Proceedings of the Twenty-Ninth AAAI Conference on Artificial Intelligence. January 25-30, 2015, Austin, Texas. New York: ACM, 2015: 4200-4201. DOI: 10.5555/2888116.2888327.
[12] Bi X, Qu A N, Wang J H, et al. A group-specific recommender system[J]. Journal of the American Statistical Association, 2017, 112(519): 1344-1353. DOI: 10.1080/01621459.2016.1219261.
[13] Wang J H. Consistent selection of the number of clusters via crossvalidation[J]. Biometrika, 2010, 97(4): 893-904. DOI: 10.1093/biomet/asq061.
[14] Miller B N, Albert I, Lam S K, et al. MovieLens unplugged: experiences with an occasionally connected recommender system[C]//Proceedings of the 8th international conference on Intelligent user interfaces. January 12-15, 2003, Miami, Florida, USA. New York: ACM, 2003: 263-266. DOI: 10.1145/604045.604094.
