A novel hierarchical approach to ranking-based collaborative filtering

Abstract

In this paper, we propose a novel recommendation method that exploits the intrinsic hierarchical structure of the item space to overcome known shortcomings of current collaborative filtering techniques. A number of experiments on the MovieLens dataset, suggest that our method alleviates the problems caused by the sparsity of the underlying space and the related limitations it imposes on the quality of recommendations. Our tests show that our approach outperforms other state-ofthe- art recommending algorithms, having at the same time the advantage of being computationally attractive and easily implementable. © Springer-Verlag Berlin Heidelberg 2013.