Detecting malicious user accounts using Canvas Fingerprint

Abstract

Online social network applications suffer from people owning bulk of fake accounts. These fake accounts cause several problems such as resource consumption and inaccurate study results. In many cases, the social network operators assign full time employees to detect these fake accounts. Many researchers have proposed methods that help detecting fake accounts in online social networks. This research paper proposes a new technique that depends on HTML Canvas Fingerprint to identify what accounts belong to the same person or entity. The methodology has been tested on a public web application and found to give promising results, especially when combined with the other techniques described in the future work.