Recommender Systems: An Experimental Comparison Of Two Filtering Algorithms

Abstract

In this work we provide a review of the experiments we conducted on two contrasting recommender systems ’ algorithms: classic Collaborative Filtering and Item-based Filtering. We discuss the results extracted from the experiments and test the validity of the claim that Item-based Filtering improves significantly on the performance of classic Collaborative Filtering. Finally, the results are compared with a smart non-personalized algorithm in order to evaluate the methods ’ usefulness.