Curriculum-Aware Contrastive Learning Bi-Directional Gated Recurrent Unit for Student Learning Course Recommendation and Classification

Student learning course recommendation is a system that recommends the appropriate next course for students based on their interests and performance in previous courses. In a course recommendation system, after a student completes a suggested course, the classification task involves predicting whether the student would recommend the course to others. However, traditional course recommendation models rely on collaborative filtering or static rules, and graph-based attention models frequently ignore curriculum structure and sequential dependencies between courses, which leads to irrelevant or misaligned recommendations that affect student engagement and academic progression and fail to reflect a student's learning path. To overcome these challenges, a Curriculum-Aware Contrastive Learning–Bi-directional Gated Recurrent Unit (CACL-Bi-GRU) is proposed for student learning course recommendation and classification to improve model performance by capturing personalized learning patterns and curriculum dependencies. The proposed CACL-Bi-GRU is utilized to capture both future and past dependencies in sequential course history and to predict the next course based on sequential student enrolment. The proposed model integrates contrastive learning to differentiate semantically relevant courses, whereas the Bi-GRU model learns forward and backward dynamics. The proposed model achieves high values for the Hit Ratio of top-K items (HR@K) at HR @10, Normalized Discounted Cumulative Gain of top-K items (NDCG@K) at NDCg@10, and Mean Reciprocal Rank (MRR) values of 91.85, 65.87, and 56.97, respectively, compared to the Contrastive Learning and Graph convolutional-based Attention Decay network (CLGADN). The proposed model output ensures that associating curriculum-aware context and sequential modeling improves both the relevance and accuracy of course recommendations.