More Effective Database Construction: Enhancing the Utility of Secure Databases

• Activity Number: 665
• Type: Contributed
• Date/Time: Thursday, August 2, 2012 : 10:30 AM to 12:20 PM
• Sponsor: Section on Government Statistics
• Author(s): Hui Xie*+ and Yi Qian
• Companies: University of Illinois and Northwestern University
• Keywords: Database ; Nonparametric methods ; Perturbation ; Privacy ; Shuffling

Abstract

We propose new data masking procedures for building secure database. The new class of procedures have a number of features that make them particularly attractive for general-purpose data masking tasks. They satisfy the low disclosure risk requirement for ideal data masking. As compared with the existing data masking methods, the new procedures substantially increase the utility of secure databases. In particular, unlike the existing approaches, the new procedures are capable of preserving important nonmonotonic relationships among attributes. Maintaining such relationships can be key to the success of database research and to the determination of the optimal level of policy, managerial and economic interventions. The proposed methods do not rely on data being constrained to be of a particular type or have a particular distributional shape. In fact, the methods provide unified, flexible and robust algorithms to mask general types of confidential variables with arbitrary distributional shapes. Our evaluations in simulation and real datasets demonstrate that the new methods perform better than the existing ones and represent a significant advancement in data masking techniques.