{"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/a-multi-task-approach-for-disentangling","title":"A Multi-Task Approach for Disentangling Syntax and Semantics in Sentence Representations","arxiv_id":"1904.01173","date":"2019-04-02","proceeding":"NAACL 2019 6","authors":["Mingda Chen","Qingming Tang","Sam Wiseman","Kevin Gimpel"],"abstract":"We propose a generative model for a sentence that uses two latent variables,\nwith one intended to represent the syntax of the sentence and the other to\nrepresent its semantics. We show we can achieve better disentanglement between\nsemantic and syntactic representations by training with multiple losses,\nincluding losses that exploit aligned paraphrastic sentences and word-order\ninformation. We also investigate the effect of moving from bag-of-words to\nrecurrent neural network modules. We evaluate our models as well as several\npopular pretrained embeddings on standard semantic similarity tasks and novel\nsyntactic similarity tasks. Empirically, we find that the model with the best\nperforming syntactic and semantic representations also gives rise to the most\ndisentangled representations.","url_abs":"http://arxiv.org/abs/1904.01173v1","url_pdf":"http://arxiv.org/pdf/1904.01173v1.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":"a-multi-task-approach-for-disentangling","repo_url":"https://github.com/mingdachen/disentangle-semantics-syntax","is_official":1,"mentioned_in_paper":1,"mentioned_in_github":0,"framework":"pytorch","reach":null}],"tasks":[{"task_slug":"disentanglement","task_name":"Disentanglement"},{"task_slug":"semantic-similarity","task_name":"Semantic Similarity"},{"task_slug":"semantic-textual-similarity","task_name":"Semantic Textual Similarity"},{"task_slug":"sentence","task_name":"Sentence"}],"methods":[],"datasets_introduced":[],"methods_introduced":[],"results":[],"syntology":{"atlas_url":"https://app.syntology.ai/?focus=1904.01173","mcp":null,"developers":"https://syntology.ai/developers"},"arxiv_metadata":null,"syntology_extracted_results":null}