{"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/pragmatically-informative-text-generation","title":"Pragmatically Informative Text Generation","arxiv_id":"1904.01301","date":"2019-04-02","proceeding":"NAACL 2019 6","authors":["Sheng Shen","Daniel Fried","Jacob Andreas","Dan Klein"],"abstract":"We improve the informativeness of models for conditional text generation\nusing techniques from computational pragmatics. These techniques formulate\nlanguage production as a game between speakers and listeners, in which a\nspeaker should generate output text that a listener can use to correctly\nidentify the original input that the text describes. While such approaches are\nwidely used in cognitive science and grounded language learning, they have\nreceived less attention for more standard language generation tasks. We\nconsider two pragmatic modeling methods for text generation: one where\npragmatics is imposed by information preservation, and another where pragmatics\nis imposed by explicit modeling of distractors. We find that these methods\nimprove the performance of strong existing systems for abstractive\nsummarization and generation from structured meaning representations.","url_abs":"http://arxiv.org/abs/1904.01301v2","url_pdf":"http://arxiv.org/pdf/1904.01301v2.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":"pragmatically-informative-text-generation","repo_url":"https://github.com/sIncerass/prag_generation","is_official":0,"mentioned_in_paper":1,"mentioned_in_github":0,"framework":"pytorch","reach":{"status":"ok"}},{"paper_slug":"pragmatically-informative-text-generation","repo_url":"https://github.com/reallygooday/60daysofudacity","is_official":0,"mentioned_in_paper":0,"mentioned_in_github":1,"framework":"pytorch","reach":{"status":"ok","spdx":"MIT"}}],"tasks":[{"task_slug":"abstractive-text-summarization","task_name":"Abstractive Text Summarization"},{"task_slug":"conditional-text-generation","task_name":"Conditional Text Generation"},{"task_slug":"data-to-text-generation","task_name":"Data-to-Text Generation"},{"task_slug":"grounded-language-learning","task_name":"Grounded language learning"},{"task_slug":"informativeness","task_name":"Informativeness"},{"task_slug":"text-generation","task_name":"Text Generation"}],"methods":[],"datasets_introduced":[],"methods_introduced":[],"results":[{"leaderboard":"/sota/data-to-text-generation-on-e2e-nlg-challenge","task":"Data-to-Text Generation","dataset":"E2E NLG Challenge","model":"S_1^R","rank_in_archive_order":1,"of":11,"metrics":{"BLEU":"68.60","CIDEr":"2.37","METEOR":"45.25","NIST":"8.73","ROUGE-L":"70.82"},"uses_additional_data":false}],"syntology":{"atlas_url":"https://app.syntology.ai/?focus=1904.01301","mcp":null,"developers":"https://syntology.ai/developers"},"arxiv_metadata":null,"syntology_extracted_results":null}