Abstract
This report outlines the research and development activities conducted under the 2024 Moodle Association of Japan (MAJ) Research and Development (R&D) Project, selected under the theme “Individualized and Time-Series Analysis of Multiple Quiz Attempts Using Generative AI” (Notification
of Acceptance: September 2024). In this project, we developed and implemented a method for automatically analyzing the highest scores and corresponding quiz questions obtained by students through multiple attempts on quizzes created with ‘STACK,’ a mathematics online assessment system integrated into the Moodle learning management system (LMS). The analysis was conducted utilizing the generative AI ‘OpenAI API.’ The project centered on two primary research and development activities. First, we designed a time-series evaluation model to assess students’ learning progress and understanding levels based on their quiz scores and the established evaluation model. This model enabled us to analyze how students improved their performance over time or where they encountered learning difficulties. Second, we developed and implemented a plugin to support feedback provision for students. This plugin leverages generative AI to automatically generate personalized feedback for each student, aiming to promote active learning behaviors and support self-improvement. Through these activities, we successfully enhanced the accuracy of learning data analysis and gained valuable insights into the effectiveness of generative AI utilization in educational settings.
