Disclosure Control by Computer Scientists: An Overview and an Application of Microaggregation to Mobility Data Anonymization
Josep Domingo-Ferrer, Michal Sramka
Computer Engineering and Mathematics, Universitat Rovira i Virgili, Tarragona, Catalonia, Spain

Privacy-preserving data mining (PPDM) is a subdiscipline of computer science which in many respects is parallel to statistical disclosure control (SDC) within statistics. We focus here on the connections between k-anonymity, a concept arisen in the PPDM community, and microaggregation, a family of methods developed within SDC. This is first discussed at a conceptual level. We then move to anonymization of mobility data, i.e. trajectories, a very dynamic area in PPDM and a completely neglected one in SDC. After that, we apply the microaggregation approach to k-anonymize real-world trajectories. We present a new distance measure for spatio-temporal data that facilitates the microaggregation process. The measure naturally considers both spatial and temporal aspects and can be fine-tuned for specific applications and instantiated with existing measures for spatial data, sequences, or time series.

The utility properties of our k-anonymization method for trajectories are as follows:

• Time information is taken into account and it is preserved;

• The original locations are preserved (no fake locations are introduced);

• The lengths of the microaggregated trajectories do not exactly match the lengths of the original trajectories, but they are strongly correlated;

• The shape of original trajectories is fairly well preserved;

• The number of discarded trajectories is much reduced in comparison to competing methods in the literature.

Regarding disclosure risk, experimental results reported indicate that our method achieves a lower re-identification probability compared to other competitors like (k,δ)-anonymity when the distortion is the same.

Keywords: Privacy-preserving data mining; Statistical disclosure control; Trajectory anonymization

Biography: Josep Domingo-Ferrer is a Full Professor of Computer Science and an ICREA-Acadèmia Research Professor at Universitat Rovira i Virgili, Tarragona, Catalonia, where he holds the UNESCO Chair in Data Privacy. He received his M. Sc. and Ph. D. degrees in Computer Science from the Autonomous University of Barcelona. He also holds an M. Sc. in Mathematics. His research interests are in data privacy, data security and cryptographic protocols, with a focus on the conciliation of individual privacy and corporate/national security. He has authored 3 patents and over 250 publications. He has received several scientific and entrepreneurial awards, among which the ICREA-Acadèmia Research Prize 2008 from the Government of Catalonia, which placed him among the 40 leading researchers in all areas in the Catalan universities.