Integrating Side Information into Collaborative Filtering Recommendation Method in Online Course Platform

To acquire non-formal education, one can access an online course platform. There are plenty of courses on those platforms, so the recommender system came up to help the user choose the one that matches their preferences. A recommender system with a collaborative filtering type is more suitable for non-formal education. Furthermore, a user might have some considerations for choosing a course. Therefore, we integrate side information into two collaborative filtering recommendation methods: Bayesian Personalized Ranking (BPR) and Singular Value Decomposition (SVD). The side information incorporated into BPR via feature augmentation, while we use the HybridSVD scheme for the SVD. We also tried to scale the rating matrix to promote the unpopular classes. The results show that the best top-N performance was achieved using the scaled HybridSVD with the course concept similarity matrix. © 2024 IEEE.