Privacy enhancing technologies for solving the privacy-personalization paradox: Taxonomy and survey

Abstract Personal data are often collected and processed in a decentralized fashion, within different contexts. For instance, with the emergence of distributed applications, several providers are usually correlating their records, and providing personalized services to their clients. Collected data include geographical and indoor positions of users, their movement patterns as well as sensor-acquired data that may reveal users' physical conditions, habits and interests. Consequently, this may lead to undesired consequences such as unsolicited advertisement and even to discrimination and stalking. To mitigate privacy threats, several techniques emerged, referred to as Privacy Enhancing Technologies, PETs for short. On one hand, the increasing pressure on service providers to protect users' privacy resulted in PETs being adopted. One the other hand, service providers have built their business model on personalized services, e.g. targeted ads and news. The objective of the paper is then to identify which of the PETs have the potential to satisfy both usually divergent - economical and ethical - purposes. This paper identifies a taxonomy classifying eight categories of PETs into three groups, and for better clarity, it considers three categories of personalized services. After defining and presenting the main features of PETs with illustrative examples, the paper points out which PETs best fit each personalized service category. Then, it discusses some of the inter-disciplinary privacy challenges that may slow down the adoption of these techniques, namely: technical, social, legal and economic concerns. Finally, it provides recommendations and highlights several research directions.