We found a match
Your institution may have access to this item. Find your institution then sign in to continue.
- Title
Multiple Attention Modules-based Knowledge Tracing.
- Authors
Kai Zhang; Zhengchu Qin; Yue Liu; Xinyi Qin
- Abstract
Knowledge tracing quantifies student's knowledge state by analyzing their interaction with exercises and predicts their future answers. This study proposes a Multiple Attention Modules-based Knowledge Tracing model to improve the representation of learning and forgetting behaviors in knowledge tracing. The proposed model employs three attention modules to shape learned and forgotten behaviors. The conceptual attention module calculates the similarity between concepts, while the state attention module measures the similarity between concept mastery states. The behavioral attention module helps the model to pay explicit attention to student's exercise interactions. To assess the effectiveness of the three attention modules on modeling performance, the study explores their impact on learning and forgetting behavior by ablating them in turn. The experimental results demonstrate that all three attention modules contribute positively to the modeling performance. In comparison with several other knowledge tracing models, the proposed model shows better performance on four real datasets.
- Subjects
RIGHT to be forgotten; ATTENTION
- Publication
IAENG International Journal of Computer Science, 2023, Vol 50, Issue 2, p633
- ISSN
1819-656X
- Publication type
Article