.png){kind=icon}
.png){kind=icon}
Return
Predicting Student Performance Based on Knowledge Characteristics and Learning Ability
.png){kind=icon}
OA
AI
DOI：10.1109/ACCESS.2025.3538700.png)
.png){kind=icon}
Original request for help.png){kind=icon}
Share.png){kind=icon}
Save
Abstract
En 中文
Knowledge tracking is a significant area of inquiry within contemporary educational research. Recent advancements in deep learning have led to the development of knowledge tracking models that demonstrate superior predictive capabilities compared to traditional approaches. In response to the challenges of limited interpretability and the complexities associated with long sequence dependencies in existing deep knowledge tracking frameworks, this study introduces a novel deep knowledge tracking model that incorporates an attention mechanism. This model utilizes the knowledge features of exercises alongside students' learning abilities as input layer feature information, enabling it to more effectively identify critical features at various temporal points throughout the learning process, ultimately enhancing predictive performance. Extensive experimentation was conducted using five authentic datasets, with results indicating that the proposed model offers a more precise evaluation of students' knowledge acquisition. Additionally, findings from ablation studies underscore the importance of incorporating practice difficulty and learning ability, which together enrich the input layer's feature information and yield a synergistic effect that positively influences the model's predictive accuracy. The outcomes of this research have the potential to facilitate personalized learning by monitoring the learning trajectories and knowledge states of online learners.
Keywords:
Educational data mining
knowledge tracing
deep learning
deep learning
attention mechanisms
attention mechanisms
intelli- gent education
intelli- gent education
Journal
IF:
3.6
Papers:
9.8W
Citations:
29.4W