A Question-centric Multi-experts Contrastive Learning Framework for Improving the Accuracy and Interpretability of Deep Sequential Knowledge Tracing Models

TL;DR

The paper tackles deep sequential KT by addressing the need to model question-level information and to provide interpretable predictions. It introduces Q-MCKT, a Question-centric Multi-experts KT framework that separately models question- and concept-level knowledge acquisition using MoE gates, and enhances question representations with fine-grained contrastive learning. An interpretable IRT-based prediction layer combines the learned acquisition states to yield transparent predictions, while a joint objective integrates KT loss, acquisition-score supervision, and contrastive learning. Experiments on four public KT datasets show consistent gains in AUC and interpretability, with analyses highlighting the roles of each component and the benefits of a question-centric, expert-guided approach for robust and comprehensible KT in education.

Abstract

Knowledge tracing (KT) plays a crucial role in predicting students' future performance by analyzing their historical learning processes. Deep neural networks (DNNs) have shown great potential in solving the KT problem. However, there still exist some important challenges when applying deep learning techniques to model the KT process. The first challenge lies in taking the individual information of the question into modeling. This is crucial because, despite questions sharing the same knowledge component (KC), students' knowledge acquisition on homogeneous questions can vary significantly. The second challenge lies in interpreting the prediction results from existing deep learning-based KT models. In real-world applications, while it may not be necessary to have complete transparency and interpretability of the model parameters, it is crucial to present the model's prediction results in a manner that teachers find interpretable. This makes teachers accept the rationale behind the prediction results and utilize them to design teaching activities and tailored learning strategies for students. However, the inherent black-box nature of deep learning techniques often poses a hurdle for teachers to fully embrace the model's prediction results. To address these challenges, we propose a Question-centric Multi-experts Contrastive Learning framework for KT called Q-MCKT. We have provided all the datasets and code on our website at https://github.com/rattlesnakey/Q-MCKT.

Figures (9)

• Figure 1: A graphical illustration of the KT problem. "" and "" denote the question is answered correctly and incorrectly and "" denotes the student doesn't get a chance to answer the question.
• Figure 2: The distribution of question interaction frequency in two real-world datasets.
• Figure 3: A visualization showcasing the overall scores given by multiple teachers.
• Figure 4: The graphical illustrations of prediction modeling in our Q-MCKT model.
• Figure 5: The distribution of question global accuracy in NIPS34 and EdNet Datasets.