Synthoid: Endpoint User Profile Control

摘要

Web tracking is a practice that has grown in scope and complexity as the Internet has expanded. In particular, advertising networks have been continually advancing web tracking in order to gather the most information on users. On the other hand, users have often reacted negatively to tracking schemes, expressing concern about their lack of control. We present Synthoid, a system which returns control to users, without a need for tracker cooperation or broad adoption. Synthoid achieves this by actively fetching content of a selected set of topics, imprinting the tracking network profile for the user with these topics. It eliminates the need for users to place trust in web trackers by directly controlling the signal trackers measure, without inconveniencing users. Additionally, as traffic is generated directly from the users' machines, Synthoid is able to function with any method of tracking, even those which use advanced techniques. We examine the effectiveness of Synthoid with real web trackers. We demonstrate that it is successfully able to imprint a target profile with low volume and when faced with significant interference. Finally, we show that it is able to entirely alter browsing profiles when run alongside user traces. The key behind Synthoid's performance is a strong, artificial, yet carefully constructed and semantically determined, signal.