{"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/latent-relational-metric-learning-via-memory","title":"Latent Relational Metric Learning via Memory-based Attention for Collaborative Ranking","arxiv_id":"1707.05176","date":"2017-07-17","proceeding":null,"authors":["Yi Tay","Anh Tuan Luu","Siu Cheung Hui"],"abstract":"This paper proposes a new neural architecture for collaborative ranking with\nimplicit feedback. Our model, LRML (\\textit{Latent Relational Metric Learning})\nis a novel metric learning approach for recommendation. More specifically,\ninstead of simple push-pull mechanisms between user and item pairs, we propose\nto learn latent relations that describe each user item interaction. This helps\nto alleviate the potential geometric inflexibility of existing metric learing\napproaches. This enables not only better performance but also a greater extent\nof modeling capability, allowing our model to scale to a larger number of\ninteractions. In order to do so, we employ a augmented memory module and learn\nto attend over these memory blocks to construct latent relations. The\nmemory-based attention module is controlled by the user-item interaction,\nmaking the learned relation vector specific to each user-item pair. Hence, this\ncan be interpreted as learning an exclusive and optimal relational translation\nfor each user-item interaction. The proposed architecture demonstrates the\nstate-of-the-art performance across multiple recommendation benchmarks. LRML\noutperforms other metric learning models by $6\\%-7.5\\%$ in terms of Hits@10 and\nnDCG@10 on large datasets such as Netflix and MovieLens20M. Moreover,\nqualitative studies also demonstrate evidence that our proposed model is able\nto infer and encode explicit sentiment, temporal and attribute information\ndespite being only trained on implicit feedback. As such, this ascertains the\nability of LRML to uncover hidden relational structure within implicit\ndatasets.","url_abs":"http://arxiv.org/abs/1707.05176v3","url_pdf":"http://arxiv.org/pdf/1707.05176v3.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":"latent-relational-metric-learning-via-memory","repo_url":"https://github.com/vanzytay/WWW2018_LRML","is_official":0,"mentioned_in_paper":0,"mentioned_in_github":0,"framework":"tf","reach":null}],"tasks":[{"task_slug":"attribute","task_name":"Attribute"},{"task_slug":"collaborative-ranking","task_name":"Collaborative Ranking"},{"task_slug":"metric-learning","task_name":"Metric Learning"},{"task_slug":"recommendation-systems","task_name":"Recommendation Systems"}],"methods":[],"datasets_introduced":[],"methods_introduced":[],"results":[{"leaderboard":"/sota/collaborative-filtering-on-movielens-1m","task":"Recommendation Systems","dataset":"MovieLens 1M","model":"LRML","rank_in_archive_order":20,"of":31,"metrics":{"HR@10":"0.7397","nDCG@10":"0.5453"},"uses_additional_data":false},{"leaderboard":"/sota/collaborative-filtering-on-movielens-20m","task":"Recommendation Systems","dataset":"MovieLens 20M","model":"LRML","rank_in_archive_order":2,"of":18,"metrics":{"HR@10":"0.8447","nDCG@10":"0.6152"},"uses_additional_data":false},{"leaderboard":"/sota/collaborative-filtering-on-netflix","task":"Recommendation Systems","dataset":"Netflix","model":"LRML","rank_in_archive_order":8,"of":10,"metrics":{"Recall@10":"0.5371","nDCG@10":"0.3578"},"uses_additional_data":false}],"syntology":{"atlas_url":null,"mcp":null,"developers":"https://syntology.ai/developers"},"arxiv_metadata":null,"syntology_extracted_results":null}