<![CDATA[The recommendation system in online learning plays a crucial role in supporting personalized and adaptive learning. However, traditional approaches often overlook the relationships between material concepts and the difficulty level of the learning content. This study proposes a recommendation model based on Graph Neural Networks (GNN), utilizing a graph representation of learning materials with difficulty level attributes. A dummy dataset with 100 materials and 200 conceptual relationships was used for testing. The evaluation results show that the proposed GNN model achieves a Precision@3 of 0.01, Recall@3 of 0.375, and NDCG@3 of 0.01, which are higher than baseline methods such as collaborative filtering and content-based filtering. This indicates that GNN can enhance the relevance of learning material recommendations. Future research will focus on using real-world datasets and exploring heterogeneous GNN models to improve recommendation performance. This model also contributes to designing a recommendation system that can adjust to the abilities and needs of learners. Future research is expected to test this model with larger and real datasets and explore the application of GNN models on heterogeneous data to examine the potential for improving recommendation performance. By considering the difficulty level of materials, this model has the potential to improve the learning experience, making it more relevant and adaptive for users.]]>
Copyrights © 2025
