{"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/fact-discovery-from-knowledge-base-via-facet","title":"Fact Discovery from Knowledge Base via Facet Decomposition","arxiv_id":"1904.09540","date":"2019-04-21","proceeding":"NAACL 2019 6","authors":["Zihao Fu","Yankai Lin","Zhiyuan Liu","Wai Lam"],"abstract":"During the past few decades, knowledge bases (KBs) have experienced rapid\ngrowth. Nevertheless, most KBs still suffer from serious incompletion.\nResearchers proposed many tasks such as knowledge base completion and relation\nprediction to help build the representation of KBs. However, there are some\nissues unsettled towards enriching the KBs. Knowledge base completion and\nrelation prediction assume that we know two elements of the fact triples and we\nare going to predict the missing one. This assumption is too restricted in\npractice and prevents it from discovering new facts directly. To address this\nissue, we propose a new task, namely, fact discovery from knowledge base. This\ntask only requires that we know the head entity and the goal is to discover\nfacts associated with the head entity. To tackle this new problem, we propose a\nnovel framework that decomposes the discovery problem into several facet\ndiscovery components. We also propose a novel auto-encoder based facet\ncomponent to estimate some facets of the fact. Besides, we propose a feedback\nlearning component to share the information between each facet. We evaluate our\nframework using a benchmark dataset and the experimental results show that our\nframework achieves promising results. We also conduct extensive analysis of our\nframework in discovering different kinds of facts. The source code of this\npaper can be obtained from https://github.com/thunlp/FFD.","url_abs":"http://arxiv.org/abs/1904.09540v1","url_pdf":"http://arxiv.org/pdf/1904.09540v1.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":"fact-discovery-from-knowledge-base-via-facet","repo_url":"https://github.com/thunlp/FFD","is_official":1,"mentioned_in_paper":1,"mentioned_in_github":1,"framework":"none","reach":null}],"tasks":[{"task_slug":"knowledge-base-completion","task_name":"Knowledge Base Completion"},{"task_slug":"relation-prediction","task_name":"Relation Prediction"}],"methods":[],"datasets_introduced":[],"methods_introduced":[],"results":[],"syntology":{"atlas_url":null,"mcp":null,"developers":"https://syntology.ai/developers"},"arxiv_metadata":null,"syntology_extracted_results":null}