Exploiting reviews to generate personalized and justified recommendations to guide users' selections

Abstract

We introduce BUS, a recommender that assists users by providing personalized and justified suggestions to facilitate the task of deciding which items, among the recommended ones, are best tailored towards their individual interests. We exploit users' reviews and matrix factorization to generate recommendations that include reviewers' opinions related to item characteristics that each individual user frequently mentions. To demonstrate the validity of RUS we use the Amazon dataset.