Article: Predicting academic performance in electromechanical courses based on a knowledge-point association graph Journal: International Journal of Information and Communication Technology (IJICT) 2026 Vol.27 No.84 pp.25 - 48 Abstract: Academic performance prediction is essential for personalised teaching intervention, yet electromechanical courses involve complex precedence dependencies and logical correlations between knowledge points that traditional models fail to capture. To address this, we propose a learning-performance prediction method named knowledge-graph-based mechatronics education learning prediction, which integrates expert knowledge with historical interaction data to construct a knowledge graph encompassing 56 core knowledge points. This method employs graph convolutional networks (GCNs) to integrate topological features for representation learning and introduces a temporal-attention mechanism to model the dynamic evolution of knowledge mastery. Experimental results demonstrate that the proposed method achieves an area under the curve of 0.894, which is 5.2% to 11.7% higher than baseline models, and a normalised discounted cumulative gain at 5 of 0.856, significantly outperforming comparative methods. The findings highlight the effectiveness of incorporating knowledge-graph structures for accurate academic performance prediction in electromechanical education. Inderscience Publishers - linking academia, business and industry through research

Title: Predicting academic performance in electromechanical courses based on a knowledge-point association graph

Authors: Dejun Miao; Rongyan Xu; Yuequan Qiu

Addresses: School of Electrical and Automotive Engineering, Yangzhou Polytechnic University, Yangzhou, 225009, China ' School of Tourism, Yangzhou Polytechnic University, Yangzhou, 225009, China ' School of Electrical and Automotive Engineering, Yangzhou Polytechnic University, Yangzhou, 225009, China

Abstract: Academic performance prediction is essential for personalised teaching intervention, yet electromechanical courses involve complex precedence dependencies and logical correlations between knowledge points that traditional models fail to capture. To address this, we propose a learning-performance prediction method named knowledge-graph-based mechatronics education learning prediction, which integrates expert knowledge with historical interaction data to construct a knowledge graph encompassing 56 core knowledge points. This method employs graph convolutional networks (GCNs) to integrate topological features for representation learning and introduces a temporal-attention mechanism to model the dynamic evolution of knowledge mastery. Experimental results demonstrate that the proposed method achieves an area under the curve of 0.894, which is 5.2% to 11.7% higher than baseline models, and a normalised discounted cumulative gain at 5 of 0.856, significantly outperforming comparative methods. The findings highlight the effectiveness of incorporating knowledge-graph structures for accurate academic performance prediction in electromechanical education.

Keywords: academic performance prediction; knowledge graph; graph convolutional network; GCN; temporal attention.

DOI: 10.1504/IJICT.2026.155481

International Journal of Information and Communication Technology, 2026 Vol.27 No.84, pp.25 - 48

Received: 24 Mar 2026
Accepted: 27 Apr 2026
Published online: 03 Aug 2026 *

Title: Predicting academic performance in electromechanical courses based on a knowledge-point association graph

Keep up-to-date