{"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/unifying-knowledge-graph-learning-and","title":"Unifying Knowledge Graph Learning and Recommendation: Towards a Better Understanding of User Preferences","arxiv_id":"1902.06236","date":"2019-02-17","proceeding":null,"authors":["Yixin Cao","Xiang Wang","Xiangnan He","Zikun Hu","Tat-Seng Chua"],"abstract":"Incorporating knowledge graph (KG) into recommender system is promising in\nimproving the recommendation accuracy and explainability. However, existing\nmethods largely assume that a KG is complete and simply transfer the\n\"knowledge\" in KG at the shallow level of entity raw data or embeddings. This\nmay lead to suboptimal performance, since a practical KG can hardly be\ncomplete, and it is common that a KG has missing facts, relations, and\nentities. Thus, we argue that it is crucial to consider the incomplete nature\nof KG when incorporating it into recommender system.\n  In this paper, we jointly learn the model of recommendation and knowledge\ngraph completion. Distinct from previous KG-based recommendation methods, we\ntransfer the relation information in KG, so as to understand the reasons that a\nuser likes an item. As an example, if a user has watched several movies\ndirected by (relation) the same person (entity), we can infer that the director\nrelation plays a critical role when the user makes the decision, thus help to\nunderstand the user's preference at a finer granularity.\n  Technically, we contribute a new translation-based recommendation model,\nwhich specially accounts for various preferences in translating a user to an\nitem, and then jointly train it with a KG completion model by combining several\ntransfer schemes. Extensive experiments on two benchmark datasets show that our\nmethod outperforms state-of-the-art KG-based recommendation methods. Further\nanalysis verifies the positive effect of joint training on both tasks of\nrecommendation and KG completion, and the advantage of our model in\nunderstanding user preference. We publish our project at\nhttps://github.com/TaoMiner/joint-kg-recommender.","url_abs":"http://arxiv.org/abs/1902.06236v1","url_pdf":"http://arxiv.org/pdf/1902.06236v1.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":"unifying-knowledge-graph-learning-and","repo_url":"https://github.com/TaoMiner/joint-kg-recommender","is_official":1,"mentioned_in_paper":0,"mentioned_in_github":0,"framework":"pytorch","reach":null}],"tasks":[{"task_slug":"graph-learning","task_name":"Graph Learning"},{"task_slug":"knowledge-graph-completion","task_name":"Knowledge Graph Completion"},{"task_slug":"recommendation-systems","task_name":"Recommendation Systems"},{"task_slug":null,"task_name":"Relation"}],"methods":[],"datasets_introduced":[],"methods_introduced":[],"results":[{"leaderboard":"/sota/knowledge-graph-completion-on-dbbook2014","task":"Knowledge Graph Completion","dataset":"DBbook2014","model":"KTUP (soft)","rank_in_archive_order":1,"of":1,"metrics":{"Hits@10":"60.75","Mean Rank":"499"},"uses_additional_data":false},{"leaderboard":"/sota/knowledge-graph-completion-on-movielens-1m","task":"Knowledge Graph Completion","dataset":"MovieLens 1M","model":"KTUP (soft)","rank_in_archive_order":1,"of":1,"metrics":{"Hits@10":"48.9","Mean Rank":"527"},"uses_additional_data":false},{"leaderboard":"/sota/recommendation-systems-on-dbbook2014","task":"Recommendation Systems","dataset":"DBbook2014","model":"KTUP (soft)","rank_in_archive_order":1,"of":2,"metrics":{"HR@10":"0.3461","NDCG":"0.2762"},"uses_additional_data":false},{"leaderboard":"/sota/collaborative-filtering-on-movielens-1m","task":"Recommendation Systems","dataset":"MovieLens 1M","model":"KTUP (soft)","rank_in_archive_order":25,"of":31,"metrics":{"HR@10":"0.8903","NDCG":"0.6992"},"uses_additional_data":false}],"syntology":{"atlas_url":"https://app.syntology.ai/?focus=1902.06236","mcp":null,"developers":"https://syntology.ai/developers"},"arxiv_metadata":null,"syntology_extracted_results":null}