Privacy preserving targeted advertising and recommendations

ABSTRACT Recommendation systems form the centerpiece of a rapidly growing trillion dollar online advertisement industry. Curating and storing profile information of users on web portals can seriously breach their privacy. Modifying such systems to achieve private recommendations without extensive redesign of the recommendation process typically requires communication of large encrypted information, making the whole process inefficient due to high latency. In this paper, we present an efficient recommendation system redesign, in which user profiles are maintained entirely on their device/web-browsers, and appropriate recommendations are fetched from web portals in an efficient privacy-preserving manner. We base this approach on precomputing compressed data structures from historical data and running low latency lookups when providing recommendations in real-time.