A hybrid recommender system for enhanced e-commerce recommendations.

Abstract

The e-commerce sector has experienced significant growth, leading industry players to implement recommendation systems aimed at enhancing user satisfaction and boosting revenue through personalized suggestions. As a result, hybrid filtering techniques combining collaborative filtering (CF) and content-based methods have emerged as a promising approach to capture user preferences effectively Our study presents a novel hybrid recommendation system for e-commerce that combines content-based filtering using inverse document frequency-term frequency (IDF-TF) with collaborative filtering through singular value decomposition (SVD). By integrating linear regression and leveraging the ? parameter, we effectively balance between contentbased and collaborative filtering to address cold start issues and data sparsity. This approach enhances recommendation accuracy and quality, as demonstrated by our experiments on real e-commerce data, focusing on scalability, accuracy, and coverage.