An experimental comparative study of web mining methods for recommender systems

Abstract

An essential goal of the present web engineering is the development of efficient and competitive applications. This objective can be achieved by building recommender systems endowed with suitable web mining algorithms. Multiclassifiers are reliable data mining models that have been hardly used in the web system area. The paper presents a comparative study among different simple classifiers and multiclassifiers using a dataset from MovieLens recommender system. The aim of the work is to identify when the use of multiclassifiers in this type of systems is efficient.