Abstract:
In online social networks, users may choose to keep a lot of important information in the private
domain. Users may choose to hide their location, gender, interests, and affiliations. With usage
of social networks and their penetration in our lives on the rise, privacy of such information is
an important concern. Now, Online social networks also exist in application ecosystems that
allow them to access data beyond their user space such as get access to phone contact lists
through mobile phone applications. We study the possibility of privacy leaks for these users
when their entire network and its links are analyzed. Important information which was meant
to be kept secret by some users can be discerned from their connections and the network. While there is a strong emphasis on individual privacy, we analyze collective privacy of a set of users and their corresponding mentions network. We discern information like location and biography by profiling users based on their connections in the mentions network. We analyze the privacy leaks on Twitter and also evaluate the predictability of information for a user by using only information from the connections which joined Twitter before our user, hence create a complete shadow profile.
