k-NMF Anonymization in Social Network Data Publishing
Social network platforms have become very popular due to their easy access and advanced features. Although social network dataset is very useful in research and market analysis, there are different risks and attacks that can breach user privacy. Online social network such as Facebook, Google Plus and LinkedIn provide a feature that allows finding out number of mutual friends (NMF) between two users. Adversary can use such information to identify individual user and related information. As the published dataset itself reveals the information of mutual friends for every connection, it becomes very easy for an adversary to re-identify users. Existing anonymization techniques for mutual friends attack are based on edge anonymization. Such methods perform anonymization operation without considering the NMF requirement of other edges that results into more edge insertion operations and low data utility of the anonymized dataset. In this paper, we propose a k-anonymization approach that works on mutual friend sequence to ensure existence of at least k elements holding the same value such that the data utility is preserved efficiently. The vertex selection process to increase the mutual friend value for one edge reduces the requirement for other edges too. The experimental results demonstrate that the proposed anonymization approach preserves the user privacy and data utility.
Conference Details :
The Computer Journal
Date :08/02/2018 ,
Venue : Journal Paper
Published At :It is journal paper ,
Link : Paper Link