Kazuhiro Minami (kminami@ism.ac.jp)
The Institute of Statistical Mathematics
Yutaka Abe
National Statistics Center
Abstract
To perform statistical disclosure control (SDC) on multiple tables is a challenging task because sensitive information can be revealed from intersections of multiple tables involv-ing a common set of variables. This task is particularly dificult when each table contains a subset of the common variables because the intersection of those tables could form a subspace of any shape in the multi-dimensional domain space. To address this issue, we extend our SDC tool for solving a cell suppression problem of a two-dimensional table to support multi-dimensional ones. Our approach is to construct a single consolidated high-dimensional table from lower-dimensional multiple linked tables so that we can rep-resent the constraints of each input table in an integrated way. Our tool detects possible sensitive cells, which could be overlooked if each table is examined separately, by solving a cell suppression problem on the multi-dimensional table. In this paper, we describe an overview of the new SDC tool and show that we eliminate the risk of unintended informa-tion disclosure on sensitive cell values in our previous implementation based on a common decomposition technique for linear programming.
Keywords: statistical disclosure control, cell suppression problem, linear programming, R.