多代理指導系統對大學程式設計課程學生自我調整學習與學習成效之影響;The Impact of a Multi-Agent Guidance System on Students’ Self-Regulated Learning and Learning Performance in a University Programming Course

NCU Institutional Repository > 資訊電機學院 > 資訊工程研究所 > 博碩士論文 > Item 987654321/98195

請使用永久網址來引用或連結此文件: https://ir.lib.ncu.edu.tw/handle/987654321/98195

題名:	多代理指導系統對大學程式設計課程學生自我調整學習與學習成效之影響;The Impact of a Multi-Agent Guidance System on Students’ Self-Regulated Learning and Learning Performance in a University Programming Course
作者:	江昱賢;Jiang, Yu-Xian
貢獻者:	資訊工程學系
關鍵詞:	生成式AI;自我調整學習;體驗式學習;程式流程圖;程式教育;AI 代理
日期:	2025-07-01
上傳時間:	2025-10-17 12:28:34 (UTC+8)
出版者:	國立中央大學
摘要:	程式設計已成為目前學生應具備的核心能力之一，但對多數初學者而言，學習程式設計的過程常面臨邏輯理解困難、錯誤難以修正與學習動機不足等問題。為協助學生強化程式邏輯與改善錯誤反思能力，本研究基於自我調整學習 (Self-Regulated Learning, SRL) 理論，結合生成式人工智慧 (Generative AI) 與多代理系統 (Multi-Agent System)，開發個人化程式流程圖輔助系統 (Personalized Code Flowchart Assistant, PCFA)。PCFA融合鷹架教學策略 (Scaffolding) 與體驗式學習理論 (Experiential Learning)，透過階層式提示與個人化流程圖生成功能，引導學生學習程式邏輯，並在錯誤中反思與修正。為驗證系統成效，本研究以非理工背景的大學生為對象，課程期間分為傳統教學組與PCFA應用組，透過程式能力測驗與自我調整學習問卷 (Online Self-Regulated Learning Questionnaire, OSLQ) 比較兩組在學習成效與策略運用上的差異。此外，也針對不同生成式AI模型所產出的流程圖進行分析，與教師繪製的流程圖進行相似度比對，以評估其作為程式學習工具的有效性。研究結果顯示，PCFA 的使用有助於提升學生的程式設計能力與自我調整學習表現，特別是在目標設定、環境結構與尋求協助三個面向表現顯著。透過流程圖等視覺化與鷹架式指導，學生能更有效地辨識並修正邏輯錯誤，促進反思與問題解決能力。本研究為程式學習工具的開發與應用提供了實證依據，也為程式教學導入視覺引導帶來更多可能性。 ;Programming is a key skill for modern learners, yet many beginners struggle with logic, error correction, and motivation. To support learning, this study developed the Personalized Code Flowchart Assistant (PCFA), a programming learning tool combining Generative AI and a Multi-Agent System. Integrating Scaffolding and Experiential Learning, the system provides tiered prompts and personalized flowcharts to enhance logical thinking and encourage reflective learning through error analysis. Participants were non-STEM university students, divided into a traditional instruction group and a PCFA-supported group. Learning outcomes and self-regulated learning strategies were assessed through programming tests and the OSLQ questionnaire. Additionally, as part of the experimental design, flowcharts generated by various generative AI models were compared to teacher-created counterparts to evaluate their structural similarity and potential educational value. The results showed that using PCFA improved students’ programming performance and self-regulated learning, particularly in the areas of goal setting, environment structuring, and help seeking. By offering visual and scaffolded guidance, PCFA helped students identify and correct logic errors, encouraging reflection and problem-solving. These findings underscore the value of personalized visual feedback in enhancing cognitive understanding and fostering self-regulated learning in programming education.
顯示於類別:	[資訊工程研究所] 博碩士論文

文件中的檔案:

檔案	描述	大小	格式	瀏覽次數
index.html		0Kb	HTML	18	檢視/開啟

在NCUIR中所有的資料項目都受到原著作權保護.

社群 sharing

資料載入中.....