Graph Neural Network Empowers Intelligent Education: A Systematic Review From an Application Perspective

Chaobo He, Jian Luo, Yong Tang, Guanliang Chen, Quanlong Guan

Abstract: Intelligent education, also known as education driven by artificial intelligence (AI), has emerged as a research hot spot. AI technologies can empower various intelligent education applications, such as learning behaviors analysis and adaptive learning. In recent years, graph neural network (GNN), as an advanced AI technology, has demonstrated its superiority in extracting insights from graph-structured data. It also has attracted significant attention in the field of intelligent education, where graph-structured data are ubiquitous (e.g., student/teacher collaboration networks). Although previous review studies have explored the applications of GNNs in some fields, there has been limited discussion on their specific applications and potential in intelligent education. To fill this gap, we collected 85 representative studies on GNN applications in intelligent education and conducted a systematical review through the bibliometric analysis. This review uncovers three key findings: 1) benefiting from the vigorous promotion by Chinese scholars, GNN applications in intelligent education have demonstrated a remarkable growth over the past six years; 2) various GNNs have been extensively utilized across six pivotal application scenarios: learner modeling, learning performance prediction, learning resource recommendation, intelligent teaching assistance, classroom behavior analysis, and educational emotion analysis, all of which have exhibited high performance; and 3) despite notable challenges, research on GNN empowering intelligent education still has many promising directions, such as enhancing interpretability, integrating multimodal educational data, collaborating with a large language model, and expanding more applications.

External IDs:doi:10.1109/tlt.2025.3630276