{"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-to-search-in-long-documents-using","title":"Learning to Search in Long Documents Using Document Structure","arxiv_id":"1806.03529","date":"2018-06-09","proceeding":"COLING 2018 8","authors":["Mor Geva","Jonathan Berant"],"abstract":"Reading comprehension models are based on recurrent neural networks that\nsequentially process the document tokens. As interest turns to answering more\ncomplex questions over longer documents, sequential reading of large portions\nof text becomes a substantial bottleneck. Inspired by how humans use document\nstructure, we propose a novel framework for reading comprehension. We represent\ndocuments as trees, and model an agent that learns to interleave quick\nnavigation through the document tree with more expensive answer extraction. To\nencourage exploration of the document tree, we propose a new algorithm, based\non Deep Q-Network (DQN), which strategically samples tree nodes at training\ntime. Empirically we find our algorithm improves question answering performance\ncompared to DQN and a strong information-retrieval (IR) baseline, and that\nensembling our model with the IR baseline results in further gains in\nperformance.","url_abs":"http://arxiv.org/abs/1806.03529v2","url_pdf":"http://arxiv.org/pdf/1806.03529v2.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-to-search-in-long-documents-using","repo_url":"https://github.com/mega002/DocQN","is_official":1,"mentioned_in_paper":1,"mentioned_in_github":1,"framework":"tf","reach":null}],"tasks":[{"task_slug":"information-retrieval","task_name":"Information Retrieval"},{"task_slug":"question-answering","task_name":"Question Answering"},{"task_slug":"reading-comprehension","task_name":"Reading Comprehension"},{"task_slug":"retrieval","task_name":"Retrieval"}],"methods":[{"method_slug":"convolution","method_name":"Convolution"},{"method_slug":"dqn","method_name":"DQN"},{"method_slug":"dense-connections","method_name":"Dense Connections"},{"method_slug":"q-learning","method_name":"Q-Learning"}],"datasets_introduced":[],"methods_introduced":[],"results":[],"syntology":{"atlas_url":"https://app.syntology.ai/?focus=1806.03529","mcp":null,"developers":"https://syntology.ai/developers"},"arxiv_metadata":null,"syntology_extracted_results":null}