{"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/representing-meaning-with-a-combination-of","title":"Representing Meaning with a Combination of Logical and Distributional Models","arxiv_id":"1505.06816","date":"2015-05-26","proceeding":"CL 2016 12","authors":["I. Beltagy","Stephen Roller","Pengxiang Cheng","Katrin Erk","Raymond J. Mooney"],"abstract":"NLP tasks differ in the semantic information they require, and at this time\nno single se- mantic representation fulfills all requirements. Logic-based\nrepresentations characterize sentence structure, but do not capture the graded\naspect of meaning. Distributional models give graded similarity ratings for\nwords and phrases, but do not capture sentence structure in the same detail as\nlogic-based approaches. So it has been argued that the two are complementary.\nWe adopt a hybrid approach that combines logic-based and distributional\nsemantics through probabilistic logic inference in Markov Logic Networks\n(MLNs). In this paper, we focus on the three components of a practical system\nintegrating logical and distributional models: 1) Parsing and task\nrepresentation is the logic-based part where input problems are represented in\nprobabilistic logic. This is quite different from representing them in standard\nfirst-order logic. 2) For knowledge base construction we form weighted\ninference rules. We integrate and compare distributional information with other\nsources, notably WordNet and an existing paraphrase collection. In particular,\nwe use our system to evaluate distributional lexical entailment approaches. We\nuse a variant of Robinson resolution to determine the necessary inference\nrules. More sources can easily be added by mapping them to logical rules; our\nsystem learns a resource-specific weight that corrects for scaling differences\nbetween resources. 3) In discussing probabilistic inference, we show how to\nsolve the inference problems efficiently. To evaluate our approach, we use the\ntask of textual entailment (RTE), which can utilize the strengths of both\nlogic-based and distributional representations. In particular we focus on the\nSICK dataset, where we achieve state-of-the-art results.","url_abs":"http://arxiv.org/abs/1505.06816v5","url_pdf":"http://arxiv.org/pdf/1505.06816v5.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":"representing-meaning-with-a-combination-of","repo_url":"https://github.com/ibeltagy/rrr","is_official":1,"mentioned_in_paper":1,"mentioned_in_github":0,"framework":"none","reach":null}],"tasks":[{"task_slug":"knowledge-base-construction","task_name":"Knowledge Base Construction"},{"task_slug":"lexical-entailment","task_name":"Lexical Entailment"},{"task_slug":"natural-language-inference","task_name":"Natural Language Inference"},{"task_slug":"rte","task_name":"RTE"},{"task_slug":"sentence","task_name":"Sentence"}],"methods":[],"datasets_introduced":[],"methods_introduced":[],"results":[],"syntology":{"atlas_url":"https://app.syntology.ai/?focus=1505.06816","mcp":null,"developers":"https://syntology.ai/developers"},"arxiv_metadata":null,"syntology_extracted_results":null}