Improving Pairwise Preference Mining Algorithms Using Preference Degrees

Abstract

Different preference mining techniques designed to predict a preference order on objects have been proposed in the literature, with very good accuracy results. In this paper, we propose to consider not only the fact that the user prefer an item i1 to an item i2 but also the degree  of his preference on the two items.  We propose the algorithm FuzzyPrefMiner designed to predict fuzzy preferences and show through a series of experiments that it outperforms  pairwise preference mining techniques whose  training phase do not include information on preference degrees.