Knowledge Tracing

Knowledge tracing (KT) aims to model student learning progress by predicting future performance based on past interactions with educational materials. Current research focuses on improving KT accuracy and interpretability through various deep learning architectures, including recurrent neural networks, graph neural networks, and the integration of large language models (LLMs) to leverage semantic information from questions and student responses. These advancements are driven by the need for more personalized and adaptive learning systems, and the resulting models offer potential for improved educational assessment and feedback mechanisms. Furthermore, research is actively addressing challenges such as data sparsity, cold-start problems, and the development of more robust and explainable models.