結合生成式人工智慧與4F動態回顧循環理論於國小閱讀學習同伴系統的應用與成效評估;Application and Effectiveness Evaluation of a Reading Learning Companion System that Integrates Generative AI with the 4Fs Active Reviewing Cycle Theory in Primary School

NCUIR > College of Electrical Engineering & Computer Science > Graduate Institute of Network Learning Technology > Electronic Thesis & Dissertation > Item 987654321/95658

Please use this identifier to cite or link to this item: http://ir.lib.ncu.edu.tw/handle/987654321/95658

Title:	結合生成式人工智慧與4F動態回顧循環理論於國小閱讀學習同伴系統的應用與成效評估;Application and Effectiveness Evaluation of a Reading Learning Companion System that Integrates Generative AI with the 4Fs Active Reviewing Cycle Theory in Primary School
Authors:	李曼綾;Lee, Man-Ling
Contributors:	網路學習科技研究所
Keywords:	生成式人工智慧;4F動態回顧循環;聊天機器人;學習同伴;Generative Artificial Intelligence;4Fs Active Reviewing Cycle;Chatbot;Reading Companion
Date:	2024-07-27
Issue Date:	2024-10-09 17:07:43 (UTC+8)
Publisher:	國立中央大學
Abstract:	在台灣，身教式持續安靜閱讀(MSSR)已廣泛實施，但由於大師生比，學生在閱讀速度以及閱讀興趣上大不相同，教師難以對每位學生的閱讀成效進行有效評估。為解決這個問題，本研究結合生成式人工智慧(GenAI)聊天機器人，以及4F動態回顧循環 (Active Reviewing Cycle)理論，開發閱讀學習同伴平台，以OpenAI的Assistants API串接，建立聊書機器人做為學生的閱讀學習同伴。此閱讀學習同伴透過檢索學生所閱讀的書籍內容以及根據學生所輸入的回應，提出4F :事實、發現、感覺及未來問題，從不同角度引導學生進行書籍內容的反思、回憶及延伸，為學生提供一個互動式的閱讀體驗，從而增強學生的閱讀理解能力。為了調查生成式AI閱讀同伴的影響及評估這種方法的有效性，本研究於台灣北部某非學校型態實驗教育機構進行為期八週的實驗，共收集了37名國小學生的閱讀理解前後測、問卷以及系統日誌資料。並建立學生的閱讀聊書互動歷程分析儀表板，以視覺化圖表方式為教師及學生提供聊書互動過程中的行為量化數據，如:聊書次數、與學習同伴互動的聊書來回數，以及在互動過程中所觸發的模組(事實、未來等)序列等資訊，為教師提供一個可以即時了解班級中所有學生聊書互動之工具。透過這種方式學生不但能即時的與學習同伴進行聊書獲得即時的反饋，也能透過觀察閱讀歷程分析儀表板上的資訊，來提升自己的閱讀及聊書策略。研究結果顯示，使用AI聊天機器人作為閱讀學習同伴可以顯著提高學生的閱讀理解能力。然而，就學生的動機而言，結果並沒有顯示出顯著的進步。在培養學生動機的能力方面，仍需進一步的策略和方法的探索。本研究所開發之閱讀互動歷程分析儀表板也為學生及教師提供了即時的反饋，教師能夠透過4F問題之回答，監控學生的閱讀聊書狀況，從而進行更有效的教學策略調整，也為學生提供一個觀察自己或同儕聊書互動的自我監控工具，以更好的調整和提升自己的聊書互動行為。綜上所述，本研究證實生成式AI聊天機器人結合4F問題在閱讀教育領域，特別是在提高國小學生閱讀理解能力方面的潛力，閱讀互動歷程儀表板也為教育者提供了一個有效的工具來監控和促進學生的閱讀進展。未來的研究可望通過改進生成式AI技術和互動設計，進一步拓展生成式AI在教育領域的應用，特別是在提升學生的動機及自我調整能力方面，使生成式AI成為更有效的學習同伴。 ;In Taiwan, the implementation of Modeled Sustained Silent Reading (MSSR) in primary schools is widespread. However, due to the high student-teacher ratio, students vary significantly in their reading speeds and interests, making it challenging for teachers to effectively assess each student′s reading outcomes. To address this issue, the study incorporates Generative Artificial Intelligence (GenAI) chatbot and the 4Fs Active Reviewing Cycle theory to develop a Reading Companion Platform. Leveraged by OpenAI′s Assistants API, the platform establishes a Book Chatbot as a reading companion for students. This companion engages students by retrieving the content of books they read and generating questions based on the 4Fs questions: Facts, Findings, Feelings, and Future, thus guiding students through reflection, recall, and extension of the book content from different perspectives, providing an interactive reading experience that enhances reading comprehension. To investigate the impact and assess the effectiveness of the GenAI reading companion, an eight-week experiment was conducted in a primary school in northern Taiwan, involving 37 primary students. The study collected data on reading comprehension pre- and post-tests, questionnaires, and system logs. A Reading Portfolio Dashboard for analyzing student-book talk interactions was also developed, providing visualized data charts to teachers and students. These charts displayed quantitative data such as the number of chat sessions, the number of interactions with the reading companion, and the sequence of triggered modules (Fact, Future, etc.) during the interactions, offering teachers a real-time tool to understand all students′ book talk interactions within the class. Through this method, students could not only talk about books with their reading companion for immediate feedback but also enhance their reading and book talking strategies by observing the data on the Reading Portfolio Dashboard. The results indicated that using an AI chatbot as a reading companion significantly improved students′ reading comprehension. However, there were no significant improvements in students′ motivation. Further exploration of strategies and methods is needed to cultivate students′ motivational capabilities. The Reading Portfolio Dashboard developed in this study provided students and teachers with immediate feedback. Teachers were able to monitor the students′ interactions during book talk interactions through responses to the 4F questions, enabling them to adjust teaching strategies more effectively. Additionally, the dashboard offered students a self-monitoring tool to observe their own or their peers′ interactions in book talk interactions, helping them improve and fine-tune their engagement in these discussions. In conclusion, this study confirms the potential of Generative AI chatbots in the field of reading education, especially in enhancing primary students′ reading comprehension. It also provides educators with an efficient tool to track and support students′ reading development. Future research could broaden the application of Generative AI in education by enhancing AI technologies and interactive designs, particularly in boosting students′ motivation and self-regulation skills, thus making Generative AI a more valuable learning companion.
Appears in Collections:	[Graduate Institute of Network Learning Technology] Electronic Thesis & Dissertation

Files in This Item:

File	Description	Size	Format
index.html		0Kb	HTML	22	View/Open

社群 sharing

Loading...