Developing a price-sensitive recommender system to improve accuracy and business performance of ecommerce applications

Much work has been done on recommender systems (RS) and much evidence was collected from applications about their effectiveness on business. As a consequence, the use of RS has quickly shifted from information retrieval to automatic marketing tools. The main aim of marketing tools is to positively affect customers' purchasing decisions and we know through marketing literature that purchasing decisions are strongly influenced by price. However, few works have explored the issue of including price in a recommendation engine. In this paper, we want to describe the main issues of designing this type of price-sensitive recommendation engine. We want also to demonstrate what the effect is of this design on recommendations' accuracy and on business performance. We demonstrate that including price in an RS improves the accuracy of recommendations, but it has to be properly modeled in order to also improve business performance. We have experimented with a Price-Sensitive RS in a laboratory setting and compared it to a traditional one by varying several settings