{"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/improving-machine-reading-comprehension-with","title":"Improving Machine Reading Comprehension with General Reading Strategies","arxiv_id":"1810.13441","date":"2018-10-31","proceeding":"NAACL 2019 6","authors":["Kai Sun","Dian Yu","Dong Yu","Claire Cardie"],"abstract":"Reading strategies have been shown to improve comprehension levels,\nespecially for readers lacking adequate prior knowledge. Just as the process of\nknowledge accumulation is time-consuming for human readers, it is\nresource-demanding to impart rich general domain knowledge into a deep language\nmodel via pre-training. Inspired by reading strategies identified in cognitive\nscience, and given limited computational resources -- just a pre-trained model\nand a fixed number of training instances -- we propose three general strategies\naimed to improve non-extractive machine reading comprehension (MRC): (i) BACK\nAND FORTH READING that considers both the original and reverse order of an\ninput sequence, (ii) HIGHLIGHTING, which adds a trainable embedding to the text\nembedding of tokens that are relevant to the question and candidate answers,\nand (iii) SELF-ASSESSMENT that generates practice questions and candidate\nanswers directly from the text in an unsupervised manner.\n  By fine-tuning a pre-trained language model (Radford et al., 2018) with our\nproposed strategies on the largest general domain multiple-choice MRC dataset\nRACE, we obtain a 5.8% absolute increase in accuracy over the previous best\nresult achieved by the same pre-trained model fine-tuned on RACE without the\nuse of strategies. We further fine-tune the resulting model on a target MRC\ntask, leading to an absolute improvement of 6.2% in average accuracy over\nprevious state-of-the-art approaches on six representative non-extractive MRC\ndatasets from different domains (i.e., ARC, OpenBookQA, MCTest, SemEval-2018\nTask 11, ROCStories, and MultiRC). These results demonstrate the effectiveness\nof our proposed strategies and the versatility and general applicability of our\nfine-tuned models that incorporate these strategies. Core code is available at\nhttps://github.com/nlpdata/strategy/.","url_abs":"http://arxiv.org/abs/1810.13441v2","url_pdf":"http://arxiv.org/pdf/1810.13441v2.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":"improving-machine-reading-comprehension-with","repo_url":"https://github.com/nlpdata/strategy","is_official":1,"mentioned_in_paper":1,"mentioned_in_github":1,"framework":"tf","reach":null}],"tasks":[{"task_slug":"arc","task_name":"ARC"},{"task_slug":"language-modeling","task_name":"Language Modeling"},{"task_slug":"language-modelling","task_name":"Language Modelling"},{"task_slug":"machine-reading-comprehension","task_name":"Machine Reading Comprehension"},{"task_slug":"multiple-choice","task_name":"Multiple-choice"},{"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-storycloze","task":"Question Answering","dataset":"StoryCloze","model":"Reading Strategies Model","rank_in_archive_order":7,"of":23,"metrics":{"Accuracy":"88.3"},"uses_additional_data":false}],"syntology":{"atlas_url":"https://app.syntology.ai/?focus=1810.13441","mcp":null,"developers":"https://syntology.ai/developers"},"arxiv_metadata":null,"syntology_extracted_results":null}