Reconstructing Profiles from Information Disseminated on the Internet

摘要

In this paper, we warn social network users about the threat that their profiles can easily be rebuilt from information disseminated on the Internet. The micro-blogging site Twitter is supposed to retain less personal information than sites like Facebook. Despite this, is it possible to reconstruct Twitter user profiles solely from publicly available information? We propose a new system based on a method of re-identification, which consists in two phases. Starting from a given Twitter user profile (first database), we try in the first phase to find his or her information scattered on other websites, such as blogs or social networks, from which we obtain a second database. The second phase of re-identification consists in forming a link between the two databases in order to reconstruct the digital identity of the user. We based our experiment on 250 randomly selected Twitter profiles on which we attempted to identify their owners. We believe that we have managed to recognize 41.6% of our sample. We conclude that the digital identity of users can be easily reconstructed solely from publicly available data that they or others have freely made available on the Internet.