A Review of Differential Privacy in Individual Data Release

The rapid development of mobile technology has improved users' quality of treatment, and tremendous amounts of medical information are readily available and widely used in data analysis and application, which bring on serious threats to users' privacy. Classical methods based on cryptography and anonymous-series models fail due to their high complexity, poor controllability, and dependence on the background knowledge of adversaries when it comes to current mobile healthcare applications. Differential privacy is a relatively new notion of privacy and has become the de facto standard for a security-controlled privacy guarantee. In this paper, the key aspects of basic concepts and implementation mechanisms related to differential privacy are explained, and the existing research results are concluded. The research results presented include methods based on histograms, tree structures, time series, graphs, and frequent pattern mining data release methods. Finally, shortcomings of existing methods and suggested directions for future research are presented.