The Use of Regression Methodology for the Compromise of Confidential Information in Statistical Databases

Abstract

A regression methodology based technique can be used to compromise confidentiality in a statistical database. This holds true even when the DBMS prevents application of regression methodology to the database. Existing inference controls, including cell restriction, perturbation, and table restriction approaches, are shown to be generally ineffective against this compromise technique. The effect of incomplete supplemental knowledge on the regression methodology based compromise technique is examined. Finally, some potential complicators of this disclosure scheme are introduced. © 1987, ACM. All rights reserved.