Personalized Recommendation Service in University Libraries using Hybrid Collaborative Filtering Recommendation System

University libraries provide a disciplinary function in addressing the educational demands of students, researchers, and faculty. However, it overloads the system with data, making it difficult for users to access relevant resources. Traditional Collaborative Filtering Recommendation (CFR) System often suffer with the data sparsity and cold-start problem which leads to low user level satisfaction. To address these issues, a hybrid CFR is proposed in this research, to improve the recommendation services in university libraries. The hybrid recommendation system combines content-based filtering and user-based collaborative filtering and the clustering algorithm. To reduce the sparsity, precise K-means clustering is used to cluster the similar users together to get the better result for the recommendations. Also, the adaption of content-based methods reduces the cold-start issue and provides a way of ensuring that new users and items are recommended the right study resources. The proposed hybrid CFR system is consequently applied for a personalized recommendation services in university library and show better recommended accuracy with 99.74%, recommended recall rate of 0.73, and Mean Square Error (MSE) of 0.49 when compared to existing CFR methods optimized hybrid CFR and Singular Value Decomposition (SVD) based CFR.