A Personalized Collaborative Recommendation Approach Based on Clustering of Customers

Abstract

Collaborative filtering has been known to be the most successful recommender techniques in recommendation systems. Collaborative methods recommend items based on aggregated user ratings of those items and these techniques do not depend on the availability of textual descriptions. They share the common goal of assisting in the users search for items of interest, and thus attempt to address one of the key research problems of the information overload. Collaborative filtering systems can deal with large numbers of customers and with many different products. However there is a problem that the set of ratings is sparse, such that any two customers will most likely have only a few co-rated products. The high dimensional sparsity of the rating matrix and the problem of scalability result in low quality recommendations. In this paper, a personalized collaborative recommendation approach based on clustering of customers is presented. This method uses the clustering technology to form the customers centers. The personalized collaborative filtering approach based on clustering of customers can alleviate the scalability problem in the collaborative recommendations.