A New Data Perturbation Method of Reference Control in Statistical Database

Abstract

Data perturbation is a method of reference control in statistical database. Data modification and data exchange are two methods of data perturbation. Data modification replaces the original data with the sample data sequence after deciding probability distribution function of the original data and generating sample data sequence. Data swapping interchanges attribute values of records. While the algorithm of the former is complex, this paper presents a new method in which the sample sequence generation is skipped. We simply sort the original data, divide them into different groups and exchange attribute values of the same group. Experiments show that the proposed method can effectively simplify the former method and guarantee the safety of the statistical data at the same time.

Randomization of Statistical Queries of Type Median: A Simulation Approach

“…Other masking techniques involve random noise addition either to the query itself or the query output (Giggins and Brankovic, 2012;Hegadi et al, 2011), while other techniques ensure perturbation via microaggregation (Muralidhar and Sarathy, 1999). Some perturbation techniques consist in swapping query elements based on distributions calculated from the database elements (Zou and Zhang, 2012).…”