- 學位論文
在程式設計課程引入基於大型語言模型的同儕推薦機制對學生學習策略影響之研究
A Study on the Influence of Student Learning Strategies by Employing LLM-based Peer Recommendation Mechanisms in Programming Courses
摘要
本研究探討在混合式學習環境的初等程式設計課程中，引入基於大型語言模型設計的同儕推薦機制對學生學習策略的影響。本研究從自行建構的程式設計教學平台上開發了一個推薦機制，此推薦機制原設計原先用於程式碼抄襲檢測，後擴展至推薦程式碼風格相近的學生之間進行互助，目的是促進同儕間的交流與互動。這對於混合式教學的線上環境中經常因缺乏同儕互動而難以尋求幫助的學生尤其重要。本研究透過MSLQ學習動機策略問卷來評估學習動機與學習策略，並以考試成績作為學習成效的衡量標準，目的是評估同儕推薦機制對學習動機策略的影響及其對學習成效的潛在貢獻。研究分析學生在多項學習策略上的表現，發現在實施推薦系統的學年與未實施推薦系統的學年相比，「尋求幫助」的策略顯示了顯著的差異。結果顯示，推薦機制的引入對於促進學生在「尋求幫助」策略上的使用頻率下降，這一變化反而對學習成效的提升表現出正面的貢獻。除此之外，本研究還探討了不同學年度的學生以及不同入學管道在混合式程式設計教育中是否存在學習動機、學習策略與學習成效的差異，以及學習動機和學習策略變化的可能原因和特定學習策略對學習成效的影響。這些發現為混合式程式設計課程以及混合式學習環境中技術工具的有效應用提供了實證支持。
並列摘要
On a self-constructed programing platform, this study investigates the impact of a peer recommendation mechanism on students’ learning strategies. The recommendation mechanism was initially designed for detecting plagiarism in code. It was later extended to recommend students with similar coding styles to aid each other. This is particularly vital in blended learning online environments where peer interaction is often limited. The study utilized the Motivated Strategies for Learning Questionnaire (MSLQ) to assess learning motivation and strategies and used examination scores as a measure of learning outcomes. The aim was to evaluate the impact of the peer recommendation mechanism on learning strategies and its potential contribution to learning outcomes. Analysis of student performance across various learning strategies indicated significant differences in the " help seeking " strategy between different academic years with and without the recommendation system. The findings support that the introduction of the recommendation mechanism significantly reduced the frequency of students' "help seeking," but positively contributing to learning outcomes. Furthermore, the study explored whether there are differences in learning motivation, strategies, and outcomes among students from different academic years and admission channels within the blended programming education. It also examined potential reasons for changes in learning motivation and strategies and the specific impact of certain learning strategies on learning outcomes. These insights offer empirical support for the effective application of technological tools in blended programming courses and blended learning environments.