ACE-KT: Cascaded Cognitive Modeling for Stage-wise Knowledge Tracing

Teng Guo, Yubin Xia, Jinsen Ke, Mingliang Hou, Jiaqi Zheng, Zitao Liu

Abstract: Knowledge Tracing (KT) aims to predict students’ academic performance by modeling their knowledge mastery over time, based on their historical learning interactions. However, current KT models often oversimplify student interactions by treating them as standard time series rather than as cognitive processes. Such loose assumptions fail to capture the complexities of cognitive evolution. Consequently, modeling student learning as a process of cognitive transformation rather than as a mere sequence of time-stamped events remains a fundamental challenge in KT research. To address this issue, we propose ACE-KT (cAscaded Cognitive modEling for Knowledge Tracing), a novel framework inspired by cognitive process theory, which shifts the focus from purely sequential modeling to cognitive representation learning. Specifically, we design a cascaded cognitive framework that simulates the human cognitive process in three sequential stages: Rhythm Perception Module implemented via a convolutional layer to extract pre-attentive patterns; Contextual Structuring Module realized through a Transformer encoder to capture contextual and relational dependencies; and Cognitive Integration Module achieved using an enhanced selective structured state space model, equipped with stacked linear transformations and TELU activations, to support temporal abstraction and cognitive consolidation. Extensive experiments on five real-world datasets demonstrate that \textbf{ACE-KT} consistently outperforms 22 state-of-the-art KT baselines, validating its effectiveness in capturing the cognitive dynamics underlying students’ cognitive processes. The source code and technical appendix are provided in the Supplementary Material.

Submission Number: 2037