Logo
ScienceStack
Table of Contents
Fetching ...
Sign In

MAGE-KT: Multi-Agent Graph-Enhanced Knowledge Tracing with Subgraph Retrieval and Asymmetric Fusion

Chi Yu, Hongyu Yuan, Zhiyi Duan

TL;DR

MAGE-KT tackles the brittleness and inefficiency of graph-based knowledge tracing by coupling a multi-agent, multi-view framework that builds a heterogeneous KC graph and a student–question interaction graph. It introduces a target-student conditioned subgraph retrieval mechanism and an asymmetric cross-attention fusion module to focus computation on high-value, KC-grounded evidence. The approach delivers state-of-the-art predictive performance across three KT datasets and yields higher-fidelity KC relations via a dedicated multi-agent extraction pipeline. This framework improves both the accuracy of next-question predictions and the quality of learned KC relations, with implications for scalable, interpretable KT in real-world educational settings.

Abstract

Knowledge Tracing (KT) aims to model a student's learning trajectory and predict performance on the next question. A key challenge is how to better represent the relationships among students, questions, and knowledge concepts (KCs). Recently, graph-based KT paradigms have shown promise for this problem. However, existing methods have not sufficiently explored inter-concept relations, often inferred solely from interaction sequences. In addition, the scale and heterogeneity of KT graphs make full-graph encoding both computationally both costly and noise-prone, causing attention to bleed into student-irrelevant regions and degrading the fidelity of inter-KC relations. To address these issues, we propose a novel framework: Multi-Agent Graph-Enhanced Knowledge Tracing (MAGE-KT). It constructs a multi-view heterogeneous graph by combining a multi-agent KC relation extractor and a student-question interaction graph, capturing complementary semantic and behavioral signals. Conditioned on the target student's history, it retrieves compact, high-value subgraphs and integrates them using an Asymmetric Cross-attention Fusion Module to enhance prediction while avoiding attention diffusion and irrelevant computation. Experiments on three widely used KT datasets show substantial improvements in KC-relation accuracy and clear gains in next-question prediction over existing methods.

MAGE-KT: Multi-Agent Graph-Enhanced Knowledge Tracing with Subgraph Retrieval and Asymmetric Fusion

TL;DR

MAGE-KT tackles the brittleness and inefficiency of graph-based knowledge tracing by coupling a multi-agent, multi-view framework that builds a heterogeneous KC graph and a student–question interaction graph. It introduces a target-student conditioned subgraph retrieval mechanism and an asymmetric cross-attention fusion module to focus computation on high-value, KC-grounded evidence. The approach delivers state-of-the-art predictive performance across three KT datasets and yields higher-fidelity KC relations via a dedicated multi-agent extraction pipeline. This framework improves both the accuracy of next-question predictions and the quality of learned KC relations, with implications for scalable, interpretable KT in real-world educational settings.

Abstract

Knowledge Tracing (KT) aims to model a student's learning trajectory and predict performance on the next question. A key challenge is how to better represent the relationships among students, questions, and knowledge concepts (KCs). Recently, graph-based KT paradigms have shown promise for this problem. However, existing methods have not sufficiently explored inter-concept relations, often inferred solely from interaction sequences. In addition, the scale and heterogeneity of KT graphs make full-graph encoding both computationally both costly and noise-prone, causing attention to bleed into student-irrelevant regions and degrading the fidelity of inter-KC relations. To address these issues, we propose a novel framework: Multi-Agent Graph-Enhanced Knowledge Tracing (MAGE-KT). It constructs a multi-view heterogeneous graph by combining a multi-agent KC relation extractor and a student-question interaction graph, capturing complementary semantic and behavioral signals. Conditioned on the target student's history, it retrieves compact, high-value subgraphs and integrates them using an Asymmetric Cross-attention Fusion Module to enhance prediction while avoiding attention diffusion and irrelevant computation. Experiments on three widely used KT datasets show substantial improvements in KC-relation accuracy and clear gains in next-question prediction over existing methods.
Paper Structure (19 sections, 6 equations, 1 figure, 3 tables)

This paper contains 19 sections, 6 equations, 1 figure, 3 tables.

Table of Contents

  1. Introduction
  2. Methodology
  3. Heterogeneous Graph Construction
  4. Multi-relational KC Graph.
  5. S–Q Interaction Graph.
  6. Multi-view Fusion Prediction
  7. Subgraph Retrieval
  8. Asymmetric Cross-Attention Fusion
  9. Prediction and Loss Function
  10. Experiments and results
  11. Datasets
  12. Baselines
  13. Implementation Details
  14. Main Results
  15. Ablation Results
  16. ...and 4 more sections

Figures (1)

  • Figure 1: The MAGE-KT framework.