主题：Collaborative Topic Ranking:Leveraging Item Meta-data for Sparsity Reduction
报告人：
Dr. Jing He is currently an associate professor in college of engineering and science at Victoria University, Australia. She has been awarded a PhD degree from Academy of Mathematics and System Science, Chinese Academy of Sciences in 2006. Prior to joining to Victoria University, she worked in University of Chinese Academy of Sciences, China during 2006-2008. She has been active in areas of Data Mining and Artificial Intelligence, Web service/Web search, Spatial and Temporal Database, Multiple Criteria Decision Making, Intelligent System, Scientific Workflow and some industry field such as E-Health,Petroleum Exploration and Development, Water recourse Management and eResearch. She has published over 60 research papers in the refereed international journals and conference proceedings including ACM transaction on Internet Technology (TOIT), IEEE Transaction on Knowledge and Data Engineering (TKDE), Information System, Plos one, The Computer Journal, Computers and Mathematics with Applications, Concurrency and Computation: Practice and Experience, International Journal of Information Technology & Decision Making, Applied Soft Computing, and ICDE, AAAI, SIGIR, WWW. She received over 1.5 million Australia dollar research funding from Australian Research Council (ARC) with ARC early career researcher award (DECRA), ARC discovery project, ARC Linkage project and National Natural Science Foundation of China (NSFC) since 2008.
学者网主页：http://www.scholat.com/jinghe
报告内容：
Pairwise ranking methods are popular for learning recommender systems from implicit feedback. They attempt to discriminate between a handful of observed items and the large set of unobserved items. In these approaches, however, user preferences and item characteristics cannot be estimated reliably due to overfitting given highly sparse data. To alleviate this problem, in this talk, I will introduce a novel hierarchical Bayesian framework which incorporates “bag-of-words” type meta-data on items into pair-wise ranking models for one class collaborative filtering. The main idea of the method lies in extending the pair-wise ranking with a probabilistic topic modeling. Instead of regularizing item factors through a zero-mean Gaussian prior, our method introduces item-specific topic proportions as priors for item factors. As a by-product, interpretable latent factors for users and items may help explain recommendations in some applications. We conduct an experimental
study on a real and publicly available dataset, and the results show that our algorithm is effective in providing accurate recommendation and interpreting user factors and item factors.
评论 0