Disjunctive Boolean Kernels-based Collaborative Filtering for top-N Item Recommendation

In many real-world recommendation tasks the available data consists only of simple interactions between users and items, such as clicks and views, called implicit feedback. In this kind of scenarios model based pairwise methods have shown of being one of the most promising approaches. In this paper, we propose a principled and efficient kernelbased collaborative filtering method for top-N item recommendation inspired by pairwise preference learning. We also propose a new boolean kernel, called Monotone Disjunctive Kernel, which is able to alleviate the sparsity issue that is one of the main problem in collaborative filtering contexts. The embedding of this kernel is composed by all the combinations of a certain degree d of the input variables, and these combined features are semantically interpreted as disjunctions of the input variables. Experiments on several CF datasets have shown the effectiveness and the efficiency of the proposed kernel-based method.