Online education demands precise knowledge state tracking for personalized learning. Current methods struggle to model both knowledge dependencies and dynamic learning patterns. This paper proposes STDG-KT, a dual graph framework integrating static knowledge relationships and temporal dynamics to address these challenges. To address these challenges, we propose STDG-KT, an integrated static–temporal graph architecture for knowledge tracing. The model constructs a Static Knowledge Graph (SKG) to provide stable concept-level structural priors and a Dynamic Interaction Graph (DIG) to update learner–concept evidence along interaction sequences. A time-aware encoder and a learnable decay gate are used to attenuate historical states according to elapsed intervals, while an adaptive fusion module balances structural priors and dynamic interaction evidence for prediction. Extensive experiments on three public datasets demonstrate STDG-KT's superior performance, achieving significant improvements in prediction accuracy while maintaining robust generalization ability for long-sequence learning scenarios.
Zhang et al. (Thu,) studied this question.
Synapse has enriched 5 closely related papers on similar clinical questions. Consider them for comparative context: