Peer-record enhanced contrastive learning knowledge tracing model

Knowledge Tracing (KT) is an important technology for personalized instruction. Its task is to automatically track changes in students’ knowledge levels over time based on their historical learning trajectories, so as to be able to accurately predict how students will perform in future learning, and thus provide appropriate learning counseling. In this paper, we propose a peer-record enhanced contrastive learning knowledge tracing model (PR-CLKT) to address the problem of existing research that focuses only on students’ own answer history to extract individual information. The model identifies students with similar response patterns by calculating the similarity scores between answer records, and inputs the answer sequences of these students as positive sample pairs into the contrastive learning framework, thus stimulating the model to learn a better quality representation of knowledge states. The experimental results show that the overall performance of the proposed PR-CLKT model is superior on the three benchmark datasets.