Deep Learning Based Knowledge Tracing: A Review of the Literature
Siyi Zhu, Wenjie Chen · 2025
This study presents new advances in knowledge tracing modeling with Deep Learning. Knowledge Tracing (KT) refers to assessing learners' mastery of knowledge points by analyzing their problem records. Now with deep learning techniques, DLKT models are well equipped to analyze students' complex learning processes. We divided the existing DLKT models into five categories: RNN-based models, attention-based models, GNN-based models, LLM-based models, and other innovative methods. This study compiles more than thirty DLKT models, compares their performance on seven commonly used datasets, and lists the test results for different metrics. We also discuss the main difficulties facing the knowledge tracing field and also predict future trends in this direction.