Incorporating Singular Value Decomposition in User-based Collaborative Filtering Technique for a Movie Recommendation System: A Comparative Study

Abstract

User-based collaborative filtering (UCF) technique is typically used to build a recommendation system (RS). A wide variety of techniques, such as matrix factorization, cosine similarity and Pearson correlation, have been proposed to improve the performance of the UCF algorithm in order to build more intelligent RSs. In this paper, we first describe the traditional UCF algorithm as the baseline; then we apply various techniques including singular value decomposition (SVD), cosine similarity, and Pearson correlation to examine and compare the performance of a small- scale movie RS. Our preliminary experimental results show that the UCF which used SVD and Pearson correlation performs better than a traditional UCF.