[{"data":1,"prerenderedAt":-1},["ShallowReactive",2],{"doc-detail-160324-en":3,"doc-seo-160324-105":30,"detail-sidebar-cat-0-en-105":91},{"code":4,"msg":5,"data":6},0,"success",{"doc_id":7,"user_id":8,"nickname":9,"user_avatar":10,"doc_module":4,"category_id":11,"category_name":12,"doc_title":13,"doc_description":14,"doc_content":15,"file_id":16,"file_url":17,"file_type":18,"file_size":19,"view_count":4,"is_deleted":4,"is_public":20,"is_downloadable":20,"audit_status":20,"page_count":21,"language":22,"language_code":23,"site_id":24,"html_lang":23,"table_of_contents":25,"faqs":26,"seo_title":27,"seo_description":14,"update_tm":28,"read_time":29},160324,687207020761,"Oliver Hayes","https://ap-avatar.wpscdn.com/davatar_155a257f0dc6eb9ab79c44ca47cae57d",8,"Research & Report","Question Generation for English Reading Comprehension Exercises using Transformers - Research paper","Secondary language education relies on reading comprehension tasks to evaluate students’ language ability, yet building high-quality items is time-intensive because passages must match known vocabulary and grammar while questions must target desired reasoning skills. This research trains a controllable transformer-based NLP model to generate multiple question types from user-specified passages. After fine-tuning, the new controls sometimes overfit or produced insufficient diversity. Adjusted outputs from an existing control became suitable for reading comprehension, and further improvements require more training data or alternative training schemes.","Question Generation for English Reading Comprehension Exercises using Transformers  \nAlexander Maas * , Kazunori D Yamada * , Toru Nagahama * , Taku Kawada *, Tatsuya Horita *  \nAbstract  \nIn secondary language education, one tool used by teachers to test students' language ability is reading comprehension. The construction of these problems can take a lot of time as the text needs to contain only the vocabulary and grammar the students know, and the questions also need to test the reasoning skills the teachers want to evaluate. To allow educators to use reading comprehension exercises more frequently, this research aims to alleviate the time constraint of creating these questions by training a controllable transformer-based natural language processing model to create questions of varying types and about a passage of text as specified by the user. After fine-tuning, the questions generated using the new controls either suffered from overfitting or from a lack of diversity between them, however the output of an existing question generation control was altered and became capable of generating questions suitable for use in reading comprehension. To improve the output of the new controls, more data could be used in the training, or an alternative training scheme would need to be utilized.  \nKeywords: Artificial Intelligence, English Language Education, Natural Language Processing, Reading Comprehension Exercises  \n1 Introduction  \nWhen learning anew language there are four main skills which curriculum are based on: reading, writing, speaking, and listening. Whereas each require their own approach to be taught effectively, there is flexibility available to educators in how they choose to teach these skills to students. One method to test students' understanding of new vocabulary and grammatical concepts is via reading comprehension. Such problems require students to demonstrate not only that they understand the individual vocabulary and grammar used in the problem but also the meaning behind the constructed sentences. Full understanding of the written text further requires students to combine all of these issues and to construct a meaningful whole [1]. It is this constructed meaningful whole that is tested by educators using reading comprehension problems.  \nAs alluded to above, the comprehension of the written word is not attributed to a single skill, such as improving students’ vocabularies, and as such there are various reasons why students may struggle to answer such questions. Even if each of the necessary skills for reading comprehension were improved individually, without sufficient practice using them all simultaneously, the students would still struggle to answer reading comprehension questions. Therefore, in  \n* Tohoku University, Sendai, Japan  \naddition to improving specific skills, students should continually practice reading comprehension exercises if the teachers wish for the students’reading comprehension ability to improve [2]. However, creating reading comprehension exercises is a very time-consuming process which most educators can only manage when creating examinations due to their already heavy workload. For Japanese teachers, who have many additional duties which often require their attention [3], there is not enough time in the day to assign to this additional exercise creation.  \nTherefore, to allow for more liberal use of reading comprehension questions, the time constraint on their construction needs to be addressed. With the development of natural language processing (NLP) models which utilise the transformer architecture [4], and their more recent iterations, such as ChatGPT developed by OpenAI [5], it has become possible to automate the construction of these questions.  \nTo allow educators to control the types of questions that will be generated, first the different types of reading comprehension questions need to be identified and the data labelled for use in the controlled generation model. This newly","cbCaivk0nCO7GSKy","https://ap.wps.com/l/cbCaivk0nCO7GSKy","pdf",1954605,1,12,"English","en",105,"# 1 Introduction\n# 2 Natural Language Processing and ChatGPT\n## 2.1 Background on NLP approaches and transformer-era models\n# 3 Dataset and question clustering\n# 4 Controllable generation architecture and fine-tuning\n# 5 Challenges and mitigation strategies\n# 6 Conclusion and future research","[{\"question\":\"Why is generating reading comprehension exercises time-consuming for educators?\",\"answer\":\"The text must be restricted to vocabulary and grammar students already know, and questions must also test specific reasoning skills. Educators face heavy workloads, leaving limited time to create these materials.\"},{\"question\":\"What is the main goal of this research?\",\"answer\":\"To reduce the time required for creating reading comprehension questions by training a controllable transformer-based NLP model that generates questions of varying types from a passage specified by the user.\"},{\"question\":\"What issues occurred after fine-tuning the controllable model?\",\"answer\":\"Generated questions sometimes suffered from overfitting or lacked diversity. The authors report that modifying an existing question-generation control improved suitability for reading comprehension.\"}]","Question Generation for English Reading Comprehension Exercises using Transformers - Research paper | PDF",1788053958,30,{"code":4,"msg":31,"data":32},"ok",{"site_id":24,"language":23,"slug":33,"title":13,"keywords":34,"description":14,"schema_data":35,"social_meta":86,"head_meta":88,"extra_data":90,"updated_unix":28},"question-generation-for-english-reading-comprehension-exercises-using-transformers-research-paper","",{"@graph":36,"@context":85},[37,54,68],{"@type":38,"itemListElement":39},"BreadcrumbList",[40,44,48,51],{"item":41,"name":42,"@type":43,"position":20},"https://docshare.wps.com","Home","ListItem",{"item":45,"name":46,"@type":43,"position":47},"https://docshare.wps.com/document/","Document",2,{"item":49,"name":12,"@type":43,"position":50},"https://docshare.wps.com/document/research-report/",3,{"item":52,"name":13,"@type":43,"position":53},"https://docshare.wps.com/document/question-generation-for-english-reading-comprehension-exercises-using-transformers-research-paper/160324/",4,{"url":52,"name":13,"@type":55,"author":56,"headline":13,"publisher":58,"fileFormat":61,"inLanguage":23,"description":14,"dateModified":62,"datePublished":62,"encodingFormat":61,"isAccessibleForFree":63,"interactionStatistic":64},"DigitalDocument",{"name":9,"@type":57},"Person",{"url":41,"name":59,"@type":60},"DocShare","Organization","application/pdf","2026-08-30",true,{"@type":65,"interactionType":66,"userInteractionCount":4},"InteractionCounter",{"@type":67},"ViewAction",{"@type":69,"mainEntity":70},"FAQPage",[71,77,81],{"name":72,"@type":73,"acceptedAnswer":74},"Why is generating reading comprehension exercises time-consuming for educators?","Question",{"text":75,"@type":76},"The text must be restricted to vocabulary and grammar students already know, and questions must also test specific reasoning skills. Educators face heavy workloads, leaving limited time to create these materials.","Answer",{"name":78,"@type":73,"acceptedAnswer":79},"What is the main goal of this research?",{"text":80,"@type":76},"To reduce the time required for creating reading comprehension questions by training a controllable transformer-based NLP model that generates questions of varying types from a passage specified by the user.",{"name":82,"@type":73,"acceptedAnswer":83},"What issues occurred after fine-tuning the controllable model?",{"text":84,"@type":76},"Generated questions sometimes suffered from overfitting or lacked diversity. The authors report that modifying an existing question-generation control improved suitability for reading comprehension.","https://schema.org",{"og:url":52,"og:type":87,"og:title":13,"og:site_name":59,"og:description":14},"article",{"robots":89,"canonical":52},"index,follow",{"doc_id":7,"site_id":24},{"code":4,"msg":5,"data":92},[93,97,101,105,110,115,120,122,127,130,134],{"id":20,"doc_module":4,"doc_module_name":46,"category_name":94,"show_sort_weight":95,"slug":96},"Story & Novel",90,"story-novel",{"id":47,"doc_module":4,"doc_module_name":46,"category_name":98,"show_sort_weight":99,"slug":100},"Literature",80,"literature",{"id":53,"doc_module":4,"doc_module_name":46,"category_name":102,"show_sort_weight":103,"slug":104},"Exam",70,"exam",{"id":106,"doc_module":4,"doc_module_name":46,"category_name":107,"show_sort_weight":108,"slug":109},5,"Comic",60,"comic",{"id":111,"doc_module":4,"doc_module_name":46,"category_name":112,"show_sort_weight":113,"slug":114},6,"Technology",50,"technology",{"id":116,"doc_module":4,"doc_module_name":46,"category_name":117,"show_sort_weight":118,"slug":119},7,"Healthcare",40,"healthcare",{"id":11,"doc_module":4,"doc_module_name":46,"category_name":12,"show_sort_weight":29,"slug":121},"research-report",{"id":123,"doc_module":4,"doc_module_name":46,"category_name":124,"show_sort_weight":125,"slug":126},9,"Religion & Spirituality",20,"religion-spirituality",{"id":125,"doc_module":4,"doc_module_name":46,"category_name":128,"show_sort_weight":125,"slug":129},"World Cup","world-cup",{"id":131,"doc_module":4,"doc_module_name":46,"category_name":132,"show_sort_weight":131,"slug":133},10,"Lifestyle","lifestyle",{"id":135,"doc_module":4,"doc_module_name":46,"category_name":136,"show_sort_weight":106,"slug":137},19,"General","general"]