Providing private recommendations on personal social networks

Abstract

Personal social networks are recently used to offer recommendations. Due to privacy concerns, privacy protection while generating accurate referrals is imperative. Since accuracy and privacy are conflicting goals, providing accurate predictions with privacy is challenging. We investigate generating personal social networks-based referrals without greatly jeopardizing users' privacy using randomization techniques. We perform real data-based trials to evaluate the overall performance of our proposed schemes. We analyze our schemes in terms of privacy and efficiency. Our schemes make it possible to generate accurate recommendations on social networks efficiently while preserving users' privacy. © Springer-Verlag Berlin Heidelberg 2010.