Revisiting Applicable and Comprehensive Knowledge Tracing in Large-Scale Data

2025-01-24

AI Summary

Addressing the longstanding trade-off among prediction accuracy, parallel training efficiency, and pedagogical interpretability in Deep Knowledge Tracing (DKT), this paper introduces DKT2—the first DKT model incorporating xLSTM. To bridge deep learning with educational measurement theory, DKT2 enhances input representations via Rasch model integration and proposes a novel Item Response Theory (IRT)-guided knowledge state disentanglement mechanism, enabling synergistic deep sequential modeling and psychometric grounding. Evaluated on three large-scale educational datasets, DKT2 consistently outperforms 17 state-of-the-art baselines, achieving significant gains in prediction accuracy and cross-dataset generalization. Its architecture supports efficient parallel training while yielding cognitively plausible, interpretable knowledge state estimates. Empirical validation in real-world intelligent tutoring systems confirms DKT2’s practical utility and deployment feasibility.

Technology Category

• Knowledge Representation and Reasoning: Knowledge Acquisition
• Machine Learning: Deep Learning Theory
• Data Mining & Knowledge Management: Representing, Reasoning, and Using Provenance, Trust

Application Category

• Semantics and Knowledge: Methods to enhance, augment, integrate or synergize semantic models such as knowledge graphs and LLMs
• Search and Retrieval-Augmented AI: Web learning to rank, online learning, and counterfactual learning for ranking
• User Modeling, Personalization and Recommendation: Explainable and interpretable methods for personalization

Abstract

Knowledge Tracing (KT) is a fundamental component of Intelligent Tutoring Systems (ITS), enabling the modeling of students' knowledge states to predict future performance. The introduction of Deep Knowledge Tracing (DKT), the first deep learning-based KT (DLKT) model, has brought significant advantages in terms of applicability and comprehensiveness. However, recent DLKT models, such as Attentive Knowledge Tracing (AKT), have often prioritized predictive performance at the expense of these benefits. While deep sequential models like DKT have shown potential, they face challenges related to parallel computing, storage decision modification, and limited storage capacity. To address these limitations, we propose DKT2, a novel KT model that leverages the recently developed xLSTM architecture. DKT2 enhances input representation using the Rasch model and incorporates Item Response Theory (IRT) for interpretability, allowing for the decomposition of learned knowledge into familiar and unfamiliar knowledge. By integrating this knowledge with predicted questions, DKT2 generates comprehensive knowledge states. Extensive experiments conducted across three large-scale datasets demonstrate that DKT2 consistently outperforms 17 baseline models in various prediction tasks, underscoring its potential for real-world educational applications. This work bridges the gap between theoretical advancements and practical implementation in KT.Our code and datasets will be available at https://github.com/codebase-2025/DKT2.

Problem

• Deep Knowledge Tracing
• Parallel Computing
• Model Interpretability

Innovation

• xLSTM
• Rasch Model
• IRT Theory