Can movies and books collaborate?: cross-domain collaborative filtering for sparsity reduction

摘要: The sparsity problem in collaborative filtering (CF) is a major bottleneck for most CF methods. In this paper, we consider novel approach alleviating the by transferring useritem rating patterns from dense auxiliary matrix other domains (e.g., popular movie website) to sparse target domain new book website). We do not require that users and items two be identical or even overlap. Based on limited ratings matrix, establish bridge between matrices at cluster-level of user-item order transfer more useful knowledge task domain. first compress into an informative yet compact pattern representation referred as codebook. Then, propose efficient algorithm reconstructing expanding perform extensive empirical tests show our method effective addressing data tasks, compared many state-of-the-art