LAUSR

In this paper we are concerned with geolocating the anonymous crowds of Dark Web forums. We do not focus on single users, but on the crowd as a whole. We work in two directions: The first idea is to exploit the time of all posts in the Dark Web forums to build profiles of the visiting crowds and to match the crowd profiles to that of users from known regions. Then, we develop a new dataset to detect the native language of the crowds to support and integrate this match. We assess the effectiveness of our methodology on the standard web and two Dark Web forums with users of known origin, and apply it to three controversial anonymous Dark Web forums. We believe that this work helps the community better understand the Dark Web from a sociological point of view and supports the investigation of authorities when the security of citizens is at stake.