ABSTRACT

Today, almost every browsing click that users make is collected by numerous trackers associated with a variety of online services (e.g., advertising networks, online social networks, e-commerce platforms). Users have often expressed concern about the lack of privacy and control over their personal data. Nonetheless, despite a substantial effort to expose and control this prevalent behavior, the reality is that users keep accepting updated online privacy policies, which in turn grant the gathering of more personal data. This project explores re-using this extensive tracking infrastructure for the benefits of both the users themselves and web services, with a goal of preventing online trolling (scenarios in which various groups deploy tactics to influence public opinion on the Internet, by leaving biased, false, misleading, and inauthentic comments, and then artificially amplifying their ratings). The project aims to show how the tracking infrastructure can be re-used as a user "fingerprint", allowing a lightweight and privacy-preserving form of identification for third-party web sites.

Intellectual Merit: In more detail, the project explores whether it is possible to utilize the ubiquitous online tracking of users for the direct benefit of the users themselves. Despite the fact that almost every browser click made over the last decade has been monitored by numerous online trackers, users often have a hard time proving their identity and uniqueness while using the Internet. On the other hand, many systems that rely upon open membership are often targets of online trolling, often via multiple-identity (Sybil) attacks. Organized trolling has become a serious problem in today?s Internet; some argue that it can have a profound impact on the society. The project is developing a system that would take direct advantage of the work online trackers do to record and interpret users' behavior. The key idea is to use the readily-available personalized content---generated by online trackers in real-time---as a means to verify an online user's uniqueness in a seamless and privacy-preserving manner. This personalized content, which would be collected by the users, stripped of identifying content, and uploaded, would be used to construct a multi-tracker vector representation of the user. The vector representation would then serve as a unique "fingerprint" of each user, making it difficult for attackers to appear as if they were many distinct users, thereby mitigating many trolling attacks.

Broader Impacts: The project has the capacity to make a significant impact by empowering the Internet community to combat the online trolling problem. The system under development will not only help numerous online communities regain trolling-free environments, but also re-use the extensive tracking infrastructure that already exists to provide tangible benefits for both end users and web sites. The PIs plan to design and disseminate personalization-based counter-trolling system as open-source software, which will enable effective trolling detection and counter-trolling methods and systems. Education is an integral part of the project, and the PIs plan to leverage existing institutional and special programs to recruit students from underrepresented groups into the project.

Please report errors in award information by writing to: awardsearch@nsf.gov.