Model of Personal Discount Sensitivity in Recommender Systems

Abstract

Recommender systems help users to encounter information or items that are of interest to them. Prior work on recommender systems has focused on eliciting preferences for items and neglected the personal traits in discount sensitivity. In this paper, we propose a recommender system that incorporates the influence of discounts. The effectiveness of the model is verified using a public retail dataset. The discount-sensitive model increased recommendation accuracy and modeling personal differences in this sensitivity further improved it. In order to specify the characteristics of discount sensitivity, the correlations between discount sensitivity and other traits of users and items are also investigated. The results show that discount sensitivity is positively correlated with item popularity and negatively correlated with persistence in purchase behaviors.