Information Privacy and Security

An Evolutionary Approach to Enhance Data Privacy

Publication Type:

Journal Article

Source:

Soft Computing - A Fusion of Foundations, Methodologies and Applications, Springer, Volume 15, Issue 7, p.1301-1311 (2011)

URL:

http://www.springerlink.com/content/42357225707508x2/

Keywords:

Information Privacy and Security; Evolutionary Algorithms

Abstract:

Dissemination of data with sensitive information about individuals has an implicit risk of unauthorized dis- closure. Perturbative masking methods propose the distor- tion of the original data sets before publication, tackling a difficult tradeoff between data utility (low information loss) and protection against disclosure (low disclosure risk).
In this paper we describe how information loss and disclosure risk measures can be integrated within an evolutionary algorithm to seek new and enhanced masking protections for continuous microdata. The proposed technique constitutes a hybrid approach that combines state-of-the-art protection methods with an evolutionary algorithm optimization. We also provide experimental results using three data sets in order to illustrate and empirically evaluate the application of this technique.

PRAM Optimization Using an Evolutionary Algorithm

Publication Type:

Book Chapter

Source:

Privacy in Statistical Databases, Springer, Number LNCS 6344, Corfú, Greece, p.97 - 106 (2010)

ISBN:

978-3-642-15837-7

Keywords:

Information Privacy and Security; Evolutionary Algorithms; Post Randomization Method; Information Loss; Disclosure Risk

Abstract:

PRAM (Post Randomization Method) was introduced in 1997 but it is still one of the least used methods in statistical categorical data protection. This fact is because of the difficulty to obtain a good transition matrix in order to obtain a good protection. In this paper, we describe how to obtain a better protection using an evolutionary algorithm with integrated information loss and disclosure risk measures to find the best matrix. We also provide experiments using a real dataset of 1000 records in order to empirically evaluate the application of this technique.

Syndicate content