Comparative analysis of matrix factorization techniques and their application to book recommendations systems

Abstract

This study presents a comparative analysis of three matrix factorization techniques—SVD, SVD++, and NMF—applied to book recommendation systems. The research investigates how each technique decomposes the user-book interaction matrix to uncover latent factors that influence book preferences. Using a standardized book rating dataset, the study evaluates the performance of these algorІТhms based on key metrics such as RMSE, MSE, MAE, FCP, and computational efficiency. By focusing on the abilІТy of these methods to model user-book relationships, this research provides insights into their practical application for improving book recommendations.