What do users like? Inferring users’ interests and preferences from their Facebook profiles

Abstract

User interest proles have become increasingly important with the advent of personalized web services and user-centric applications. These proles have been traditionally generated using the user’s web search history, browsing and interaction history with portals and websites. However, a social networking site such as Facebook, that provides users with a rich and powerful platform for specifying their interests and preferences unequivocally, has not been explored largely yet. In this paper, we present a novel unsupervised algorithm and system to infer an individual user’s interests and preferences from his Facebook prole data. We perform extensive evaluation of our approach with a dataset of 488 active Facebook users and demonstrate that a user’s interests can be estimated reasonably well from his/her Facebook profile.