A Statistical Approach to the Detection of Behavioral Tracking on the Web

Abstract

Online Behavioral Targeting and Tracking are controversial practices for which rigorous detection and analysis is challenging. Detecting Behavioral Targeting allows us to make claims about the practice of collecting and storing information about the activities and identities of Internet users, because access to this information is a prerequisite to Behavioral Targeting. The capacity to make strong claims about Behavioral Targeting and data collection in “the wild” would be valuable for policy makers. In this paper we present a conception of browser-server interactions and a novel statistical approach to detecting Behavioral Targeting and Tracking that leverages this formulation. This approach allows us to make precise claims about these practices and achieve valuable automation of analysis.