Secure re-publication of dynamic big data

Abstract

Dynamic data re-publication is now an emerging issue in data publishing due to the awareness of privacy disclosure in data sharing. Existing models such as k-anonymity and (-diversity only aim to provide data protection for single release. In practical, new data arrive continuously and up-to-date dataset should be released from time to time. The release of multiple anony-mized datasets (microdata) allows the attackers to learn extra knowledge by cross examines the releases within a targeted timeframe. In this paper, we study the data re-publication of dynamic big data based on the m-invariance model. In particular, we reconstruct the existing model to support re-publication for big data. We consider re-publication with insertion, deletion and update of the existing records. Counterfeit records will be used to maintain the update pattern of all the releases and to increase the false information in the knowledge learned by the attacker from the released microdata. © Springer International Publishing Switzerland 2013.