| 研究課題/領域番号 |
21K19824
|
|---|
| 研究種目 |
挑戦的研究(萌芽)
|
| 配分区分 | 基金 |
|---|
| 審査区分 |
中区分62:応用情報学およびその関連分野
|
|---|
| 研究機関 | 京都大学 |
|---|
研究代表者 |
|
|---|
| 研究分担者 |
久富 望 京都大学, 教育学研究科, 助教 (70825992)
|
|---|
| 研究期間 (年度) |
2021-07-09 – 2024-03-31
|
| 研究課題ステータス |
交付 (2022年度)
|
| 配分額 *注記 |
6,370千円 (直接経費: 4,900千円、間接経費: 1,470千円)
2022年度: 2,600千円 (直接経費: 2,000千円、間接経費: 600千円)
2021年度: 3,770千円 (直接経費: 2,900千円、間接経費: 870千円)
|
|---|
| キーワード | Modality analysis / Learning process / Learning Analytics / Recommendation / Knowledge Map / Modality Analysis / Learning Process |
|---|
| 研究開始時の研究の概要 |
Learners often get stuck or fail a task during the process of learning a new skill or knowledge. This research investigates a novel analysis method from multi source data of learner modality combined with reading behavior and knowledge mapping to predict the plateau in the mastery of prerequisites.
|
| 研究実績の概要 |
Based on the analysis of pen strokes conducted in the first year of this project, we focused on trying to identify possible problems that students were facing by having them self-explain the knowledge and skills using in the process of answering quiz questions. Students reviewed their pen stroke answers and self-explained at appropriate intervals, which generated annotated time series data of the answering process. This data was then analyzed to using NLP methods to generate sample self-explanations that contained the required knowledge components to solve the quiz item. The results were disseminated as an international conference paper. A self-explanation feedback and recommender system has been developed and implemented with results currently under analysis.
|
| 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
The project is proceeding as planned due to reduced restricts from COVID. However, some areas of investigation into self-explanation that were not previously considered have been identified, and more in-depth research will be conducted as a result.
|
| 今後の研究の推進方策 |
Results of analysis from the first (pen stroke) and second (self-explanation) year are being combined to provide overall feedback interface to students, with the possibility of recommending learning tasks to overcome problems encountered in the answering process. More in-depth investigation into automatically analyzing self-explanations using state-of-the-art NLP methods should enable fine grained feedback and possible hints into problems that have been encountered. We are also currently preparing to publish the results of this project in international journals to disseminate the findings to the boarder research community.
|
