Enhancing Cognitive Diagnosis by Modeling Learner Cognitive Structure State

TL;DR

This work targets improving cognitive diagnosis by explicitly modeling both knowledge state ($KS$) and knowledge structure state ($KUS$). It introduces CSCD, an edge-feature-based graph attention framework that jointly updates KS and KUS via DRC/URC modules implemented with EGAT, followed by a fusion step to obtain a holistic cognitive-structure representation $h^s$ used by an MLP predictor. Extensive experiments on ASSISTments2017, Junyi, and NIPS34 show CSCD achieving higher AUC, ACC, and lower RMSE than strong baselines, with ablation confirming the value of combining KS and KUS. The approach enhances predictive accuracy and interpretability, providing finer-grained insights into learners’ cognitive structures and enabling more targeted instructional interventions.

Abstract

Cognitive diagnosis represents a fundamental research area within intelligent education, with the objective of measuring the cognitive status of individuals. Theoretically, an individual's cognitive state is essentially equivalent to their cognitive structure state. Cognitive structure state comprises two key components: knowledge state (KS) and knowledge structure state (KUS). The knowledge state reflects the learner's mastery of individual concepts, a widely studied focus within cognitive diagnosis. In contrast, the knowledge structure state-representing the learner's understanding of the relationships between concepts-remains inadequately modeled. A learner's cognitive structure is essential for promoting meaningful learning and shaping academic performance. Although various methods have been proposed, most focus on assessing KS and fail to assess KUS. To bridge this gap, we propose an innovative and effective framework-CSCD (Cognitive Structure State-based Cognitive Diagnosis)-which introduces a novel framework to modeling learners' cognitive structures in diagnostic assessments, thereby offering new insights into cognitive structure modeling. Specifically, we employ an edge-feature-based graph attention network to represent the learner's cognitive structure state, effectively integrating KS and KUS. Extensive experiments conducted on real datasets demonstrate the superior performance of this framework in terms of diagnostic accuracy and interpretability.

Figures (6)

• Figure 1: The illustration of (a) A simple example of cognitive diagnosis; (b) The importance of the knowledge structure state(KUS).
• Figure 2: Structure of CSCD framework.
• Figure 3: EGAT
• Figure 4: Ablation Experiment Results.
• Figure 5: KUS: Predecessor-Successor and KUS: Dependency Relationships.