| 研究課題/領域番号 |
21K17865
|
|---|
| 研究種目 |
若手研究
|
| 配分区分 | 基金 |
|---|
| 審査区分 |
小区分62030:学習支援システム関連
|
|---|
| 研究機関 | 立命館大学 |
|---|
研究代表者 |
YAN YU 立命館大学, 情報理工学部, 助教 (30875894)
|
|---|
| 研究期間 (年度) |
2021-04-01 – 2025-03-31
|
| 研究課題ステータス |
交付 (2023年度)
|
| 配分額 *注記 |
3,640千円 (直接経費: 2,800千円、間接経費: 840千円)
2023年度: 1,040千円 (直接経費: 800千円、間接経費: 240千円)
2022年度: 1,040千円 (直接経費: 800千円、間接経費: 240千円)
2021年度: 1,560千円 (直接経費: 1,200千円、間接経費: 360千円)
|
|---|
| キーワード | Programming Education / NLP / Automatic Commenting / Automatic Grading / Machine Learning / Recurrent Neural Network / Random Forest / E-learning / Programming education / Deep Learning |
|---|
| 研究開始時の研究の概要 |
The proposal of the research combines SKP (Syntactic Knowledge Point) technology with high-performance deep learning technology to extract features representing each beginner programmer's weaknesses in program concepts.
|
| 研究実績の概要 |
In AY2023, the data collected from last year was analyzed in different ways. Feedbacks from ChatGPT students' programs were compared with human evaluations. The analysis results show that GhatGPT gives more positive feedback compared with human evaluations. These results also show that GhatGPT feedback can not give solutions to a defective student program. Several language processing models were explored to give better feedback. However, it is still ongoing.
|
| 現在までの達成度 (区分) |
現在までの達成度 (区分)
2: おおむね順調に進展している
理由
In AY2023, although there were no publishing outcomes from the research, the project ran smoothly and it is in the last stage now.
|
| 今後の研究の推進方策 |
In AY2024, I will find more collaborators to work with for the natural language processing part. In addition, I am planning to write a journal paper to publish the data analysis results and the neural network training results.
|
