A Large-Scale Empirical Study of Internet Users' Privacy Leakage in China

In recent years, information leakage incidents occur and have involved hundreds of millions of accounts on Tencent QQ. Despite recent efforts to study privacy leakages in online social networks (OSNs) such as Facebook and Twitter, little attention has been given to evaluating the risk of the leaked datasets. In this paper, we developed several methods that can successfully correlate the leaked datasets and accurately learn millions of users' confidential information such as real-name, OSN ID, age, birth date, social connection, and education background, etc. This privacy leakage is possible even when a large number of users' attributes such as real-name and age are intentionally set wrongly or not directly observable in the leaked datasets.