Online Program

Return to main conference page
Friday, May 18
Computational Statistics
Bayesian Computations and Applications
Fri, May 18, 10:30 AM - 12:00 PM
Grand Ballroom E

Masking Data Using an Entropy Approach (304357)

*Kurt Pflughoeft, University of Wisconsin Milwaukee 
Ehsanolah Soofi, University of Wisconsin at Milwaukee 
Refik Soyer, George Washington University 

Keywords: Data confidentiality, Disclosure risk, Data utility, Maximum entropy, Kullback Leibler information

Preserving confidentiality of individuals in data disclosure is of a prime concern for public and private organizations. The main challenge in data disclosure problem is to release data such that misuse by intruders is avoided while providing useful information to legitimate users for analysis. We propose an information theoretic approach for the data disclosure problem. The proposed information approach consists of developing a maximum entropy (ME) model based on essential features of the actual data, testing the adequacy of the ME model, producing disclosure data from the ME model and quantifying the discrepancy between the actual and the disclosure data. Several examples are shown using both univariate and multivariate data.