| The online behavior of users is generating an unprecedented wealth of information, which is implicitly public. First, the amount of information is increasing because users spend more time and engage more through posts, comments etc. Second, the users volunteer the information from the most part, but what is less obvious is that the ability to quickly collect, aggregate and analyze that information can provide very involved information about the user. Individual piece of information is harmless, but once combined with others could lead to revelations that the user has not imagined.;The focus of our research could be captured in the following question: how much user-specific information can be extracted from users' online behavior? We study this problem in two different domains. First, we examine commenting platforms, and examine how much information about users' can be extracted from their posts. Commenting platforms are companies that facilitate the backend management of comments for a wide range of websites. Second, we study the third-party tracking which is an Internet-wide privacy issue that happen across multiple domains/websites. Third-party tracking described the phenomenon when the web-browsing behavior of the user is revealed and collected by websites that the user has not explicitly visited.;Our contribution of this dissertation is focusing on (a) understanding how much user information can be extracted from commenting platforms, and (b) quantifying the extent of the third-party tracking. First, we develop a systematic way to profile users based on their commenting behavior. Our goal is to identify a small set of features that can profile users and reveal fundamental patterns and anomalies. Second, we propose a systematic method to detect and classify antisocial behavior in commenting platforms. We use the term antisocial behavior to describe activities like trolling, spamming, fanatic posting etc. The focus here is in developing a comprehensive and interpretable way to define and detect antisocial behaviors. Last, we quantify the extent of third party tracking and develop a user-implementable counter-measure. |
