Overview of Collaborative Filtering Techniques
Zhang Fan · 2011
Aiming at solving the problem of information overload in many areas,collaborative filtering(CF),as one of the most important techniques for building personalized recommender system,has three main categories: memory-based,model-based,and hybrid CF algorithms.In this paper,the basic procedure of memory-based CF is analyzed,four classic model-based CF algorithms and three categories of hybrid CF are introduced,and experimentation of CF algorithm is studied in three aspects: construction of action dataset,data partition,and evaluation metrics,then the problems we will meet when using CF for constructing recommender system and the performance of several classic CF algorithms are summarized.The recommender performance of each CF algorithm may be right and wrong,but the key of building a recommender system is choosing reasonable algorithm and evaluation metrics meeting the demands of recommender task and the characteristics of dataset.