Testing variants of minimum distance controlled tabular adjustment

Abstract

Controlled tabular adjustment (CTA), and its minimum distance variants, is a recent methodology for the protection of tabular data. Given a table to be protected, the purpose of the method is to fi nd the closest one that guarantees the confi dentiality of the sensitive cells. This is achieved by adding slight adjustments to the remaining cells, preferably excluding total ones, whose values are preserved. Unlike other approaches, this methodology can effi ciently protect large tables of any number of dimensions and structure. In this work, we test some minimum distance variants of CTA on a close-to-real data set, and analyze the quality of the solutions provided. As another alternative, we suggest a restricted CTA (RCTA) approach, where adjustments are only allowed in a subset of cells. This subset is a priori computed, for instance by a fast heuristic for the cell suppression problem. We discuss benefi ts of RCTA, and suggest several approaches for its solution.