Strategies for Mining User Preferences in a Data Stream Setting

Abstract

In this article, we formally introduce the problem of mining contextual preferences in a data stream setting. Contextual Preferences have been recently treated in the literature  and some methods for mining this special kind of preference have been proposed in the batch setting. Besides the formalization of the contextual preference mining problem in the stream setting, we propose two strategies for solving this problem. In order to evaluate our proposals, we implemented one of these strategies, the greedy one, and we compared its performance with a well known baseline in the literature, showing its efficiency through a set of experiments over real data.