{"about":{"site":"https://codewithpapers.app","non_affiliation":"Code with Papers and Syntology are not affiliated with, endorsed by, or sponsored by Papers with Code, Meta, or the pwc-archive mirror.","licence":"CC BY-SA 4.0","licence_url":"https://creativecommons.org/licenses/by-sa/4.0/legalcode","attribution":"https://codewithpapers.app/attribution","modified":"archive material modified by Syntology; see the attribution page"},"url":"/paper/mv-rnn-a-multi-view-recurrent-neural-network","title":"MV-RNN: A Multi-View Recurrent Neural Network for Sequential Recommendation","arxiv_id":"1611.06668","date":"2018-11-20","proceeding":null,"authors":["Cui Qiang","Wu Shu","Liu Qiang","Zhong Wen","Wang Liang"],"abstract":"Sequential recommendation is a fundamental task for network applications, and\nit usually suffers from the item cold start problem due to the insufficiency of\nuser feedbacks. There are currently three kinds of popular approaches which are\nrespectively based on matrix factorization (MF) of collaborative filtering,\nMarkov chain (MC), and recurrent neural network (RNN). Although widely used,\nthey have some limitations. MF based methods could not capture dynamic user's\ninterest. The strong Markov assumption greatly limits the performance of MC\nbased methods. RNN based methods are still in the early stage of incorporating\nadditional information. Based on these basic models, many methods with\nadditional information only validate incorporating one modality in a separate\nway. In this work, to make the sequential recommendation and deal with the item\ncold start problem, we propose a Multi-View Recurrent Neural Network (MV-RNN})\nmodel. Given the latent feature, MV-RNN can alleviate the item cold start\nproblem by incorporating visual and textual information. First, At the input of\nMV-RNN, three different combinations of multi-view features are studied, like\nconcatenation, fusion by addition and fusion by reconstructing the original\nmulti-modal data. MV-RNN applies the recurrent structure to dynamically capture\nthe user's interest. Second, we design a separate structure and a united\nstructure on the hidden state of MV-RNN to explore a more effective way to\nhandle multi-view features. Experiments on two real-world datasets show that\nMV-RNN can effectively generate the personalized ranking list, tackle the\nmissing modalities problem and significantly alleviate the item cold start\nproblem.","url_abs":"http://arxiv.org/abs/1611.06668v2","url_pdf":"http://arxiv.org/pdf/1611.06668v2.pdf","source":{"archive":"pwc-archive (Hugging Face), CC BY-SA 4.0","snapshot":"2025-07-28","licence_url":"https://creativecommons.org/licenses/by-sa/4.0/legalcode","row_kind":"abstracts"},"code_links":[{"paper_slug":"mv-rnn-a-multi-view-recurrent-neural-network","repo_url":"https://github.com/cuiqiang1990/MV-RNN","is_official":1,"mentioned_in_paper":1,"mentioned_in_github":0,"framework":"none","reach":null}],"tasks":[{"task_slug":"collaborative-filtering","task_name":"Collaborative Filtering"},{"task_slug":"sequential-recommendation","task_name":"Sequential Recommendation"}],"methods":[],"datasets_introduced":[],"methods_introduced":[],"results":[],"syntology":{"atlas_url":null,"mcp":null,"developers":"https://syntology.ai/developers"},"arxiv_metadata":null,"syntology_extracted_results":null}