Graph Continual Learning for Incremental Cognitive Diagnosis

Cognitive diagnosis (CD) aims to model students’ evolving cognitive states by learning representations of students, exercises, and knowledge concepts. Graph-based CD models (Graph-CDMs) enhance representation learning by capturing high-order relationships within interaction graphs. However, existing Graph-CDMs struggle with challenges arising from data distribution shifts and cascading effects in dynamic graph scenarios, limiting their real-world applicability. To tackle these limitations, we propose Continual Learning for Incremental Cognitive Diagnosis (GRACE), a continual learning framework for dynamic Graph-CDMs. GRACE disentangles node embeddings into stage-unique and cross-stage-generic components to localize updates and suppress graph cascade effects. To address distributional shifts, it incorporates a multi-level graph prompt mechanism—injecting node-, structure-, and temporal-level prompts—to guide adaptation while considering knowledge retention across stages. Extensive experiments on three real-world datasets demonstrate the superior performance, robustness, and adaptability of GRACE compared to traditional full-training and fine-tuning baselines across various Graph-CDMs.

Authors

Institutions

Publication Details

Journal
ACM Transactions on Intelligent Systems and Technology
Published
2026-11-21
DOI
https://doi.org/10.1145/3864050
Primary Topic
Intelligent Tutoring Systems and Adaptive Learning
Type
article
Field-Weighted Citation Impact
0.00
Controls
|||
ALL TIME
JAN
FEB
MAR
APR
MAY
JUN
JUL
AUG
SEP
OCT
NOV
article

Graph Continual Learning for Incremental Cognitive Diagnosis

Fei Liu, Le Wu, Shengwei Ji, Haowei Zhou et al.
ACM Transactions on Intelligent Systems and Technology
Intelligent Tutoring Systems and Adaptive Learning
article

Graph Continual Learning for Incremental Cognitive Diagnosis

Fei Liu, Le Wu, Shengwei Ji, Haowei Zhou, Chenao Xia, Yonghui Yang, Xin Li, Tong Xie
article en

Abstract

Cognitive diagnosis (CD) aims to model students’ evolving cognitive states by learning representations of students, exercises, and knowledge concepts. Graph-based CD models (Graph-CDMs) enhance representation learning by capturing high-order relationships within interaction graphs. However, existing Graph-CDMs struggle with challenges arising from data distribution shifts and cascading effects in dynamic graph scenarios, limiting their real-world applicability. To tackle these limitations, we propose Continual Learning for Incremental Cognitive Diagnosis (GRACE), a continual learning framework for dynamic Graph-CDMs. GRACE disentangles node embeddings into stage-unique and cross-stage-generic components to localize updates and suppress graph cascade effects. To address distributional shifts, it incorporates a multi-level graph prompt mechanism—injecting node-, structure-, and temporal-level prompts—to guide adaptation while considering knowledge retention across stages. Extensive experiments on three real-world datasets demonstrate the superior performance, robustness, and adaptability of GRACE compared to traditional full-training and fine-tuning baselines across various Graph-CDMs.

ACM Transactions on Intelligent Systems and Technology
University of Science and Technology of China (CN), Hefei University of Technology (CN), National University of Singapore (SG), Hefei University (CN), Artificial Intelligence in Medicine (Canada) (CA)
Intelligent Tutoring Systems and Adaptive Learning
AI Navigator

Ask Laika to Summarize, Analyze, and Connect papers live on the map.

Summarize Papers & Methodologies

Extract key findings, datasets, and comparative methods across publications.

Benchmark Rankings & Visual Analytics

Rank top research institutions, authors, funders, topics, and journals by Field-Weighted Citation Impact (FWCI) and paper volume with instant charts.

Connect Distant Disciplines

Bridge topological clusters on the map to find hidden collaborative intersections.

Laika access coming soon
Graph Continual Learning for Incremental Cognitive Diagnosis — Fei Liu, Le Wu, et al. · ACM Transactions on Intelligent Systems and Technology (2026) | TGRS Research Map | TGRS