{"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/textual-entailment-with-structured-attentions","title":"Textual Entailment with Structured Attentions and Composition","arxiv_id":"1701.01126","date":"2017-01-04","proceeding":"COLING 2016 12","authors":["Kai Zhao","Liang Huang","Mingbo Ma"],"abstract":"Deep learning techniques are increasingly popular in the textual entailment\ntask, overcoming the fragility of traditional discrete models with hard\nalignments and logics. In particular, the recently proposed attention models\n(Rockt\\\"aschel et al., 2015; Wang and Jiang, 2015) achieves state-of-the-art\naccuracy by computing soft word alignments between the premise and hypothesis\nsentences. However, there remains a major limitation: this line of work\ncompletely ignores syntax and recursion, which is helpful in many traditional\nefforts. We show that it is beneficial to extend the attention model to tree\nnodes between premise and hypothesis. More importantly, this subtree-level\nattention reveals information about entailment relation. We study the recursive\ncomposition of this subtree-level entailment relation, which can be viewed as a\nsoft version of the Natural Logic framework (MacCartney and Manning, 2009).\nExperiments show that our structured attention and entailment composition model\ncan correctly identify and infer entailment relations from the bottom up, and\nbring significant improvements in accuracy.","url_abs":"http://arxiv.org/abs/1701.01126v1","url_pdf":"http://arxiv.org/pdf/1701.01126v1.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":"textual-entailment-with-structured-attentions","repo_url":"https://github.com/kaayy/structured-attention","is_official":1,"mentioned_in_paper":1,"mentioned_in_github":0,"framework":"torch","reach":null}],"tasks":[{"task_slug":"natural-language-inference","task_name":"Natural Language Inference"},{"task_slug":null,"task_name":"Relation"}],"methods":[],"datasets_introduced":[],"methods_introduced":[],"results":[],"syntology":{"atlas_url":"https://app.syntology.ai/?focus=1701.01126","mcp":null,"developers":"https://syntology.ai/developers"},"arxiv_metadata":null,"syntology_extracted_results":null}