應用生成式模型輔助問題生成學習系統於國小社會 課程之研究;Applying Generative Modeling to Assist Problem Generation Learning System in National Elementary School Social Studies Curriculum

NCUIR > College of Electrical Engineering & Computer Science > Graduate Institute of Network Learning Technology > Electronic Thesis & Dissertation > Item 987654321/95671

Please use this identifier to cite or link to this item: http://ir.lib.ncu.edu.tw/handle/987654321/95671

Title:	應用生成式模型輔助問題生成學習系統於國小社會課程之研究;Applying Generative Modeling to Assist Problem Generation Learning System in National Elementary School Social Studies Curriculum
Authors:	洪丞威;Hung, Cheng-Wei
Contributors:	網路學習科技研究所
Keywords:	生成式 AI;批判性思考;問題生成;社會科教育;學習動機;Generative AI;Critical Thinking;Question Generation;Social Studies Education;Learning Motivation
Date:	2024-07-29
Issue Date:	2024-10-09 17:08:16 (UTC+8)
Publisher:	國立中央大學
Abstract:	社會課程在基礎教育中至關重要。然而，現今的社會課程主要還是藉由紙本教材搭配影片等多媒體的形式進行授課，無法有效提升學生學習動機。因此，本研究探討生成式輔助問題生成學習系統來開發一個虛擬同伴系統，旨在增強社會教育中的學習成效與學習動機。透過自行開發的考卷怪獸線上學習平台，本研究引入學習同伴的機制，學生可以領養一個小怪獸作為學習的夥伴，並且在學習中照顧與升級怪獸。系統的主要學習模式透過讓學生自由使用平板電腦拍攝學習主題相關的文字內容，系統從圖像中的重點文字進行影像辨識的擷取，並透過生成式AI產生問題，並結合遊戲化元素進行與同儕創造之問題進行練習挑戰。在此種科技輔助之下本研究結合問題創造的學習模式，學生可以從教課書尋找重點，藉由生成式模型的創造力，建構一個問題的範本，最後讓學生修改創造出屬於自己的問題。本研究也讓教師參與了整個問題創造的學習情境，教師可以透過本研究開發的學生學習歷程總覽給予學生問題修改的建議與回饋，幫助學生提高問題的品質與學習策略。本研究實驗對象為三十七名四至六年級學生，他們使用平板電腦參加了為期 10 週的社會學習課程。研究採用量化和質化資料收集方法，以確保全面瞭解學生的觀點，研究工具包括社會科能力測驗、學習動機量表、批判性思考量表、自主學習量表。採用問卷調查評估學生對系統功能的看法以及學習方法對學習動機的潛在影響。結果表明，學生的學習成就、學習動機、批判性思考以及自主學習都有顯著提高。透過分群演算法也發現越常使用本系統進行練習挑戰的學生，他們的學習成就會高於不常使用系統進行練習的學生。根據年級的分組檢定中也發現年級越高的學生在進行挑戰系統時也有著更高的答對率。本研究設計的問題創造學習模式與學習平臺提升了學生學習的效果，並找出與學習表現幅度較大的學生在學習行為上的差異。未來可以嘗試在其他領域使用此種學習模式進行學習，也可擴大實驗規模，進一步確認實驗的有效性。 ;The social studies curriculum plays a critical role in foundational education. However, the current social studies curriculum, delivered via traditional media like paper textbooks and videos, often fails to significantly boost students′ motivation to learn. Therefore, this study investigates the development of a virtual peer system by using a Generative AI question generating learning system, which aims to enhance the learning effectiveness and motivation in social education. Through the self-developed Exam Monster online learning platform, this study introduces the mechanism of learning peers, where students can adopt a small monster as a learning partner, and take care of and upgrade the monster during the learning process. The main learning mode of the system allows students to capture textbook content on their tablets, generate questions from the key words in the images, and combine them with gamification elements to practice challenges with peer-created questions. Leveraging this technology, the study integrates a problem generation learning model that enables students to identify central ideas in the textbook, craft a template for the problem using generative modeling creativity, and then alter it to formulate their unique problems. The study also allowed teachers to participate in the whole problem creation learning situation, and teachers could give students suggestions and feedback on problem modification through the overview of students′ learning history developed in this study, helping students to improve the quality of their problems and learning strategies. The research involved 37 students from fourth to sixth grade who engaged in a 10-week social studies curriculum utilizing tablet computers. Quantitative and qualitative data collection methods were used to ensure a comprehensive understanding of students′ perspectives. Instruments used in the study included the Social Studies Competency Test, Motivation to Learn Scale, Critical Thinking Scale, and Self-Directed Learning Scale. A questionnaire was used to assess the students′ perceptions of the system′s functioning and the potential impact of learning methods on learning motivation. The findings indicated that there was a significant improvement in students′ academic success, motivation, critical thinking abilities, and capacity for self-directed learning. The clustering analysis indicated that high-performing students tackled questions with greater diligence within the Paper Monster challenge platform, achieving a higher accuracy rate and frequently revisiting the challenges. Furthermore, assessments segmented by grade level showed that students at more advanced grade levels exhibited increased engagement rates in response to utilizing the challenge system. In this study, the developed question-creation learning model and platform improved student learning results and pinpointed behavioral differences among students with higher performance levels. Moving forward, we might consider applying this educational model to different fields of study. Additionally, enlarging the scope of the experiment could help us to more conclusively verify its effectiveness.
Appears in Collections:	[Graduate Institute of Network Learning Technology] Electronic Thesis & Dissertation

Files in This Item:

File	Description	Size	Format
index.html		0Kb	HTML	16	View/Open

社群 sharing

Loading...