矩阵分解技术应用到推荐系统解读.pptVIP

Matrix Factorization techniques for recommender systems Catalogue Paper Background Introduction Recommender Systems Strategies Matrix Factorization Methods A Basic Matrix Factorization Model Learning Algorithm Adding Biases Adding Input Sources Temporal Dynamics Input With Varying Confidence Levels Netflix Prize Competition Conclusion * * 1、Paper Background * * 1.Yehuda Koren,Yahoo Research 2.Robert Bell and Chris Volinsky,ATT Labs-Research 3.Paper published by the IEEE Computer Society in August 2009 4.Author won the grand Netflix Prize Competition in September 2009 2、Introduction Modern consumers are inundated with choices. More retailor have become interested in RS,which analyze patterns of user interest in products to pride personalized recommendations that suit a users taste. Netflix and A have made RS a salient part of their websites. Particularly userful for entainment products such as movies,music,and TV shows. 3、Recommender System Strategies Content Filtering Collaborative Filtering 1.Neighborhood methods ?user-oriented ?item-oriented 2.Latent Fator Model * * 3.1、Contents Filtering Create a profile for each user or product to characterize its nature. Need to gather external information. A known successful realization of content filtering is the Music Genome Project,which is used for the Internet radio service P. * * 3.2、Collaborative Filtering Analyze relationships between users and interdep- encies among products to identify new user-item as- Socitions. Disadvantages: cold start Two primary areas: neighborhood methods user-oriented item-oriented Latent factor models * * 3.2.1、Neighborhood methods Centered on computing the relationships between items or users. The item-oriented approach evaluates a user’s preference for an item based on ratings of “neighboring” items by the same user. The user-oriented approach identifies like-minded users who ca