{"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/learning-recurrent-span-representations-for","title":"Learning Recurrent Span Representations for Extractive Question Answering","arxiv_id":"1611.01436","date":"2016-11-04","proceeding":null,"authors":["Kenton Lee","Shimi Salant","Tom Kwiatkowski","Ankur Parikh","Dipanjan Das","Jonathan Berant"],"abstract":"The reading comprehension task, that asks questions about a given evidence\ndocument, is a central problem in natural language understanding. Recent\nformulations of this task have typically focused on answer selection from a set\nof candidates pre-defined manually or through the use of an external NLP\npipeline. However, Rajpurkar et al. (2016) recently released the SQuAD dataset\nin which the answers can be arbitrary strings from the supplied text. In this\npaper, we focus on this answer extraction task, presenting a novel model\narchitecture that efficiently builds fixed length representations of all spans\nin the evidence document with a recurrent network. We show that scoring\nexplicit span representations significantly improves performance over other\napproaches that factor the prediction into separate predictions about words or\nstart and end markers. Our approach improves upon the best published results of\nWang & Jiang (2016) by 5% and decreases the error of Rajpurkar et al.'s\nbaseline by > 50%.","url_abs":"http://arxiv.org/abs/1611.01436v2","url_pdf":"http://arxiv.org/pdf/1611.01436v2.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":"learning-recurrent-span-representations-for","repo_url":"https://github.com/asadovsky/nn","is_official":0,"mentioned_in_paper":0,"mentioned_in_github":1,"framework":"tf","reach":{"status":"ok"}},{"paper_slug":"learning-recurrent-span-representations-for","repo_url":"https://github.com/shimisalant/RaSoR","is_official":0,"mentioned_in_paper":0,"mentioned_in_github":1,"framework":"none","reach":{"status":"unanswered"}}],"tasks":[{"task_slug":"answer-selection","task_name":"Answer Selection"},{"task_slug":"extractive-question-answering","task_name":"Extractive Question-Answering"},{"task_slug":"natural-language-understanding","task_name":"Natural Language Understanding"},{"task_slug":"question-answering","task_name":"Question Answering"},{"task_slug":"reading-comprehension","task_name":"Reading Comprehension"}],"methods":[],"datasets_introduced":[],"methods_introduced":[],"results":[{"leaderboard":"/sota/question-answering-on-squad11","task":"Question Answering","dataset":"SQuAD1.1","model":"RaSoR (single model)","rank_in_archive_order":157,"of":213,"metrics":{"EM":"70.849","F1":"78.741"},"uses_additional_data":false},{"leaderboard":"/sota/question-answering-on-squad11-dev","task":"Question Answering","dataset":"SQuAD1.1 dev","model":"RASOR","rank_in_archive_order":43,"of":55,"metrics":{"EM":"66.4","F1":"74.9"},"uses_additional_data":false}],"syntology":{"atlas_url":"https://app.syntology.ai/?focus=1611.01436","mcp":null,"developers":"https://syntology.ai/developers"},"arxiv_metadata":null,"syntology_extracted_results":null}