{"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/specialising-word-vectors-for-lexical-1","title":"Specialising Word Vectors for Lexical Entailment","arxiv_id":"1710.06371","date":"2017-10-17","proceeding":null,"authors":["Ivan Vulić","Nikola Mrkšić"],"abstract":"We present LEAR (Lexical Entailment Attract-Repel), a novel post-processing\nmethod that transforms any input word vector space to emphasise the asymmetric\nrelation of lexical entailment (LE), also known as the IS-A or\nhyponymy-hypernymy relation. By injecting external linguistic constraints\n(e.g., WordNet links) into the initial vector space, the LE specialisation\nprocedure brings true hyponymy-hypernymy pairs closer together in the\ntransformed Euclidean space. The proposed asymmetric distance measure adjusts\nthe norms of word vectors to reflect the actual WordNet-style hierarchy of\nconcepts. Simultaneously, a joint objective enforces semantic similarity using\nthe symmetric cosine distance, yielding a vector space specialised for both\nlexical relations at once. LEAR specialisation achieves state-of-the-art\nperformance in the tasks of hypernymy directionality, hypernymy detection, and\ngraded lexical entailment, demonstrating the effectiveness and robustness of\nthe proposed asymmetric specialisation model.","url_abs":"http://arxiv.org/abs/1710.06371v2","url_pdf":"http://arxiv.org/pdf/1710.06371v2.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":"specialising-word-vectors-for-lexical-1","repo_url":"https://github.com/nmrksic/lear","is_official":1,"mentioned_in_paper":1,"mentioned_in_github":1,"framework":"tf","reach":{"status":"ok","spdx":"Apache-2.0"}}],"tasks":[{"task_slug":"lexical-entailment","task_name":"Lexical Entailment"},{"task_slug":null,"task_name":"Relation"},{"task_slug":"semantic-similarity","task_name":"Semantic Similarity"},{"task_slug":"semantic-textual-similarity","task_name":"Semantic Textual Similarity"}],"methods":[],"datasets_introduced":[],"methods_introduced":[],"results":[],"syntology":{"syntology_url":"https://syntology.ai/paper/1710.06371","atlas_url":"https://app.syntology.ai/?focus=1710.06371","mcp":null,"developers":"https://syntology.ai/developers"},"arxiv_metadata":null,"syntology_extracted_results":null}