Adnostic: Privacy Preserving Targeted Advertising

摘要

Online behavioral advertising (OBA) refers to the practicernof tracking users across web sites in order to infer userrninterests and preferences. These interests and preferencesrnare then used for selecting ads to present to the user. Therernis great concern that behavioral advertising in its presentrnform infringes on user privacy. The resulting public debatern- which includes consumer advocacy organizations,rnprofessional associations, and government agencies - isrnpremised on the notion that OBA and privacy are inherentlyrnin conflict.rnIn this paper we propose a practical architecture thatrnenables targeting without compromising user privacy. Behavioralrnprofiling and targeting in our system takes place inrnthe user's browser. We discuss the effectiveness of the systemrnas well as potential social engineering and web-basedrnattacks on the architecture. One complication is billing; adnetworksrnmust bill the correct advertiser without knowingrnwhich ad was displayed to the user. We propose an efficientrncryptographic billing system that directly solves the problem.rnWe implemented the core targeting system as a Firefoxrnextension and report on its effectiveness.