Collaborative filtering: theoretical positions and a research agenda in marketing

Abstract

The internet offers many choices of products, services and content. But the multitude of choices has made it more difficult for customers to find quickly what they are looking for. Collaborative Filtering (CF), or recommender system based-CF, is a methodology designed to perform such a recommendation task. These systems allow users to use expressed preferences of thousands of other people to find the product they desire, based on the level of similarity between tastes. The concept has emerged from convergent research on search browsers, intelligent agents and data mining, and it permits to escape the difficult question of why consumers prefer a particular product or brand. Furthermore, CF is open for the end-user and allows customers to discover things within an information environment that they probably never would have discovered otherwise. On a more practical perspective, CF through the internet allows us to focus exclusively on the similarity of preferences without social contamination: the consumer obtains recommendations to purchase a given product or brand on the basis of his or her own past preferences and on the basis of the preferences of a large group of anonymous consumers. In this paper, we will review the current state of research in consumer behaviour that provides the theoretical foundations underlying collaborative filtering. Then we will propose a research agenda in marketing keeping in mind the perspective of users i.e. consumers or marketers.