A Personalized Course Recommendation System has been Developed Using xDeepFM and RoBERTa Models

Our study introduces a personalized course recommendation system that combines the xDeepFM and RoBERTa models. It aims to assist new students on MOOC platforms in managing the immense amount of information available and enhancing their learning experience. Our system effectively matches students with courses that fit their interests, abilities, majors, prior knowledge backgrounds, and learning preferences. We use the xDeepFM to model the correlation between students and classes through explicit higher-order features to achieve this. Additionally, we utilize the RoBERTa to capture course text information. Furthermore, we incorporate an attention network layer into the model to learn the significance of different semantic features. Through comparative experiments, our study shows that our model outperforms current algorithms with an AUC score of 0.963. Our main contribution is integrating RoBERTa to provide a semantic representation of course material features and an attention layer that allows for a better understanding of course topics. Ultimately, our study aims to establish an efficient framework for personalized course recommendations on MOOC platforms by effectively integrating relationships to obtain high-quality representations of students and courses, leading to better suggestions.