期刊名称：International Journal of Computer Trends and Technology
电子版ISSN：2231-2803
出版年度：2017
卷号：44
期号：1
页码：24-28
DOI：10.14445/22312803/IJCTT-V44P104
出版社：Seventh Sense Research Group
摘要：In customary CF strategies, just the criticism network, which contains express input or understood criticism on the things given by clients, is utilized for preparing and forecast. Because of its fruitful application in recommender framework, community oriented sifting (CF) has turned into a hot examination subject in information mining and data recovery. Normally, the input grid is extras, which implies that most clients collaborate with thing. Because of this sparcity issue, customary CF just criticism lattice is scanty, which implies that most clients connect with couple of things. As of late, may specialists have proposed to use assistant data, for example, thing content, tp ease the information sparcity issue in CF. cooperative point regression(CTR) is one of the strategies which has accomplished promising execution by effectively incorporating both input data and thing content data. I numerous genuine application, other than the criticism and thing content data, there may exist relations among the things which can be useful for proposal. In this paper, we build up a novel various leveled Bayesian model called Relational Collaborative Topic Regression (RCTR), which amplifies CTR via consistently incorporating client thing input data, thing content data, and system structure among things into the same model. Probes certifiable datasets demonstrate that our model can accomplish preferred forecast exactness over the best in class strategies with lower experimental preparing time. In addition, RCTR can learn great interpretable idle stricter which are valuable for proposal.
关键词：Topic models; Collaborative filtering; recommender system; social network
