Private targeted advertising using homomorphic encryption

Abstract

Privacy of the user data in online setup is an important feature. The users of Internetdo not want to share their private information or interests with anybody.The targeted advertising domain clearly tracks the user activity by their browsinghistory or any other technique to do the necessary match-making to target correctadvertisements. The uesrs of the Internet have been using AdBlock or relatedsoftwares to opt out of the tracking/targeting process.In this work, we have applied the private information retrieval phenomenon toaddress the privacy issue in the domain of online targeted advertising. We are nottrying to replace the existing system, but rather suggesting a way around to preserveprivacy of the user and still target them with relevant advertisements. Wehave used a variant of homomorphic encryption to carry out operations, whichallows us to perform operations on encrypted data.A modified encryption scheme is proposed by mitigating the observed attack.Security analysis has been provided to support the security of proposed scheme.A new system of targeted advertising with underlying proposed scheme has beenincluded. We have performed basic experiments using the dummy data of integers,to present the response time of the proposed system. Communication cost,i.e. size of data transfer is very small which makes it feasible for real-world system.Response time achieved is also comparable with real-world scenario whichcan further be improved by setting up the system in distributed manner.