Research on Student Behavior Analysis and Personalized Recommendation in Online Education Platforms Based on Graph Neural Networks

The development of online education leads to an abundance of student learning behavior data for educational platforms. Traditional recommenders generally ignore the association relationship between students’ behavior learning when dealing with student learning behavior data, resulting in the low precision of recommender. In this paper, we utilize Graph Neural Network (GNN) to excavate student's behavior when learning in online education, and proposed a GNN-based personalized recommend model. We performed experiments on the small course dataset CourseCX-S, which involves course information, students’ information, student-course interactions. The experimental results show that in comparison with traditional recommend algorithms, the proposed GNN model is able to better find potential relationships between students and courses, hence result in better personalization and accuracy for a recommend system.