The data of social networks contains a large amount of personal information, of which may be public or insignificant, but some may be sensitive and private. Once the user privacy leaked, it may bring a variety of troubles for users. Holders of social networking data first conduct anonymization before the data is published. However, the simple anonymity method does not play a very good protection. At present, a number of de-anonymization attacks for the data releasing of social networks have arisen. These de-anonymization attacks are mostly based on the network structure with the use of feature matching and other methods. In this paper, we model the social network as a Structure-Attribute (SA) framework which integrates the structural characteristics of social networks and social node properties, adding some attribute nodes to the social network. The similarity measurement of social network nodes is proposed, with the consideration of the structure similarity and attribute similarity. The accuracy of node matching and de-anonymization is greatly improved. We conduct our de-anonymization based on the realistic dataset to verify the accuracy and efficiency.
Jiang, H., Yu, J., Hu, C., Zhang, C., & Cheng, X. (2018). SA Framework based De-anonymization of Social Networks. In Procedia Computer Science (Vol. 129, pp. 358–363). Elsevier B.V. https://doi.org/10.1016/j.procs.2018.03.089