申論題練習系統
講者/Speaker： 鄭志鵬, Cheng Chih-Peng
本研究開發了一套「申論題練習系統」 ，旨在解決科學教育中學生科學論述能力不足 ，以及教師因批改作業耗時費神而難以推動大量申論練習的教學痛點 。本系統以 Google Sheets 作為後端題庫 ，並透過 AI API（於系統中使用 Gemini 2.5 Flash 模型）與 Google Apps Script 技術 ，建構出能讓學生與 AI 進行即時、持續互動的對話前端界面 。系統亦融入遊戲化獎勵機制（如發放不同等級與造型的徽章） ，以激勵學生反覆挑戰與修正論述 ，同時將互動歷程完整記錄於後端供教師追蹤學生答題過程 。
教學實踐結果顯示，在「光學」實驗描述的申論題中，有 63.6% 的學生自發進行二次以上的修正作答 ；而在「密度建模」問題中，多次作答比例更達 85.2% ，且重複作答的學生中有 78.3% 的分數獲得顯著進步 。質化成效方面，多數學生能在單堂課內藉由 AI 的協助，自發且持續地豐富論述內容並提升申論品質 。
This study develops an "Essay Question Practice System" to address key pedagogical pain points in science education: students' lack of scientific argumentation skills and the immense, exhausting time teachers spend grading essay-type responses. Utilizing the AI API (powered by the Gemini 2.5 Flash model) alongside Google Sheets as a backend question bank and Google Apps Scripts , the system provides a frontend interface where students can interact with the AI in real time to iteratively refine their scientific arguments. Additionally, a gamification reward mechanism featuring diverse tier badges is introduced to motivate continuous challenges , while the full interaction logs are stored for teachers to review the students' learning process.
Teaching practice results show that in the essay questions describing the "optics" experiment, 63.6% of students spontaneously revised their answers more than twice ; in the "Density Modeling" task, the multiple-attempt rate reached 85.2% , with 78.3% of those repeating students showing significant score improvements. Qualitative results confirm that the majority of students successfully enriched their descriptions and enhanced the overall quality of their scientific writing within a single class period through AI assistance.