A graph theoretic linkage attack on microdata in a metric space
Martin Kroll(a),(*)
Transactions on Data Privacy 8:3 (2015) 217 - 243
Abstract, PDF
(a) University of Duisburg-Essen, Lotharstrasse 65, D-47057 Duisburg.
e-mail:martin.kroll @uni-due.de
|
Abstract
Certain methods of analysis require the knowledge of the spatial distances between entities whose data are stored in a microdata table. For instance, such knowledge is necessary and sufficient to perform data mining tasks such as nearest neighbour searches or clustering. However, when inter-record distances are published in addition to the microdata for research purposes, the risk of identity disclosure has to be taken into consideration. In order to tackle this problem, we introduce a flexible graph model for microdata in a metric space and propose a linkage attack based on realistic assumptions of a data snooper's background knowledge. This attack is based on the idea of finding a maximum approximate common subgraph of two vertex-labelled and edgeweighted graphs. By adapting a standard argument from algorithmic graph theory to our setup, this task is transformed to the maximum clique detection problem in a corresponding product graph. A toy example and experimental results show that publishing even approximate distances could increase the risk of identity disclosure unreasonably. We will concentrate on the perturbation of the distances; the anonymization of the vertex labels will play only a minor role in our simulations. Since the current version of our attack is not scalable, it can be launched only on datasets of sizes up to few thousands records. In the future we intend to explore possible ways of pushing further the limits of our approach.
|