Publication Type

Conference Proceeding Article

Publication Date

3-2014

Abstract

The wealth of information contained in online social networks has created a demand for the publication of such data as graphs. Yet, publication, even after identities have been removed, poses a privacy threat. Past research has suggested ways to publish graph data in a way that prevents the re-identification of nodes. However, even when identities are effectively hidden, an adversary may still be able to infer linkage between individuals with sufficiently high confidence. In this paper, we focus on the privacy threat arising from such link disclosure. We suggest L-opacity, a sufficiently strong privacy model that aims to control an adversary’s confidence on short multiedge linkages among nodes. We propose an algorithm with two variant heuristics, featuring a sophisticated look-ahead mechanism, which achieves the desired privacy guarantee after a few graph modifications. We empirically evaluate the performance of our algorithm, measuring the alteration inflicted on graphs and various utility metrics quantifying spectral and structural graph properties, while we also compare them to a recently proposed, albeit limited in generality of scope, alternative. Thereby, we demonstrate that our algorithms are more general, effective, and efficient than the competing technique, while our heuristic that preserves the number of edges in the graph constant fares better overall than one that reduces it.

Keywords

Algorithms, Experimentation, Theory

Discipline

Databases and Information Systems | Theory and Algorithms

Research Areas

Data Management and Analytics

Publication

Advances in Database Technology: EDBT 2014, 17th International Conference on Extending Database Technology, Athens, Greece, March 24-28, Proceedings

First Page

583

Last Page

594

ISBN

9783893180653

Identifier

10.5441/002/edbt.2014.52

Publisher

OpenProceedings

City or Country

Konstanz

Embargo Period

7-11-2017

Copyright Owner and License

Authors

Creative Commons License

Creative Commons Attribution-Noncommercial-No Derivative Works 4.0 License
This work is licensed under a Creative Commons Attribution-Noncommercial-No Derivative Works 4.0 License.

Additional URL

http://doi.org/10.5441/002/edbt.2014.52

Share

COinS