A Comprehensive Hybrid Implicit and Explicit Item‐Based Collaborative Filtering Approach with Bayesian Personalized Ranking for Enhancing Book Recommendations
Adidam Surekha, Radhika Gouni, Satya Keerthi Gorripati, Venubabu Rachapudi, S Anjali Devi, Anupama Angadi · 2025
Due to the rise of Amazon, Netflix, YouTube, and other web services, recommender systems have become more prominent daily. From e-services (recommend items to buyers that might seek their attention) to social advertisements (recommend customers the preferable content, by matching their interests), recommender systems are today inevitable in our everyday lives. There exist many e-services for which personalization permits spare time. This paper demonstrates a recommender system using Bayesian Personalized Ranking (BPR) with an emphasis on ranking the books. It is a prediction model exploited both on the ratings (E-explicit) or the user's browsing history (I-implicit) latent factors. This model can avoid the impact of sparsity and cold-start issues by integrating a generative artificial intelligence (GAI) approach. Hence, this model attains the effect of more accurate suggestions of products. Finally, based on the Book-Crossings dataset, the results of EIBTR, conventional machine learning, and matrix factorization models are compared. It is shown that EIBTR has a higher accuracy.