A Dual-Graph Convolutional Knowledge Tracking Method Based On Hidden Knowledge Extraction.

Saved in:

Bibliographic Details
Title:	A Dual-Graph Convolutional Knowledge Tracking Method Based On Hidden Knowledge Extraction.
Authors:	Feng, Wenfang¹ 1036784024@qq.com, Cao, Hang² 3118700173@qq.com, Zhao, Minrui³ 2904104078@qq.com, Xia, Zhiyuan³ xia15094905773@163.com
Source:	IAENG International Journal of Computer Science. Jul2026, Vol. 53 Issue 7, p2852-2865. 14p.
Subjects:	Hypergraphs, Graph neural networks, Learning analytics, Deep learning, Machine learning
Abstract:	Knowledge tracking (KT) is a methodology that leverages students' past responses to assess their current mastery of knowledge and predict their future proficiency in related concepts and exercises. However, existing KT approaches face several challenges, including inadequate representation of knowledge concepts and exercises, as well as insufficient consideration of the multiple relationships between them. To address these issues, this paper proposes a novel Dual-Graph Convolutional Knowledge Tracking method based on Hidden Knowledge Extraction (DGHKE). Specifically, multivariate relational information is modeled using concept-exercise association graphs and exercise-exercise association graphs, constructed via hypergraph and directed graph representations to facilitate effective representation learning. Furthermore, the HKIE-LSTM attention mechanism is employed to capture deep knowledge information potentially overlooked by graph convolution, integrating it with original features to enhance the representation of both concepts and exercises and mitigate the sparsity problem commonly observed in KT data. Online supervised models then exploit the fused information to enable mutual learning of representations, while additional supervised signals provide feedback to further improve prediction accuracy. Extensive experiments on three benchmark datasets, compared against seven state-of-the-art baselines, demonstrate that DGHKE consistently achieves superior performance. [ABSTRACT FROM AUTHOR]
	Copyright of IAENG International Journal of Computer Science is the property of International Association of Engineers (IAENG) and its content may not be copied or emailed to multiple sites without the copyright holder's express written permission. Additionally, content may not be used with any artificial intelligence tools or machine learning technologies. However, users may print, download, or email articles for individual use. This abstract may be abridged. No warranty is given about the accuracy of the copy. Users should refer to the original published version of the material for the full abstract. (Copyright applies to all Abstracts.)
Database:	Engineering Source

Description
Abstract:	Knowledge tracking (KT) is a methodology that leverages students' past responses to assess their current mastery of knowledge and predict their future proficiency in related concepts and exercises. However, existing KT approaches face several challenges, including inadequate representation of knowledge concepts and exercises, as well as insufficient consideration of the multiple relationships between them. To address these issues, this paper proposes a novel Dual-Graph Convolutional Knowledge Tracking method based on Hidden Knowledge Extraction (DGHKE). Specifically, multivariate relational information is modeled using concept-exercise association graphs and exercise-exercise association graphs, constructed via hypergraph and directed graph representations to facilitate effective representation learning. Furthermore, the HKIE-LSTM attention mechanism is employed to capture deep knowledge information potentially overlooked by graph convolution, integrating it with original features to enhance the representation of both concepts and exercises and mitigate the sparsity problem commonly observed in KT data. Online supervised models then exploit the fused information to enable mutual learning of representations, while additional supervised signals provide feedback to further improve prediction accuracy. Extensive experiments on three benchmark datasets, compared against seven state-of-the-art baselines, demonstrate that DGHKE consistently achieves superior performance. [ABSTRACT FROM AUTHOR]
ISSN:	1819656X