Deployment and evaluation of a usage based collaborative filtering recommendation system with blacklists

Abstract

Social web sites, like Palco Principal (PP), have considerably sized databases with the most varied types of information from user subscription information to user interactions. All that information is susceptible of being processed using data mining algorithms in order to extract knowledge relevant for the business. This document describes the implementation and evaluation of a system that uses data mining techniques, and that has been developed with the specific intent of generating recommendations, namely music recommendations, to users of the PP web site. The implemented system uses collaborative filtering techniques to recommend music additions to a user’s playlist. The impact of the deployed system has been evaluated online.