Injecting utility into anonymized datasets

By Daniel Kifer, Johannes Gehrke. SIGMOD Conference, 2006

338 citations recorded by the source corpus.

View Paper: https://www.semanticscholar.org/paper/a35d571de4b58b9f0ad804074eb0f20a151e5097

How researchers see this paper

The researchers whose trusted networks rank it highest, strongest advocate first. Every researcher gets a different order.

Citations in the corpus

Anonymizing Social Networks (2007, background): “There has been considerable follow-on work on algorithms for constructing k-anonymous tables [6, 12]; on improving the utility of the anonymized data [11, 13]; and on subtler aspects of disclosure, such as inferring properties of the target even without perfect re-identification [16, 17].”
Publishing Search Logs—A Comparative Study of Privacy Guarantees (2009, background): “As notable diﬀerences, Brickell et al. [8] measure the utility with respect to data mining tasks and Kifer and Gehrke [19] develop speciﬁc techniques to boost utility with respect to log-linear models.”
Fast Data Anonymization with Low Information Loss (2007, methodology): “In our experiments, we use KL Divergence, which has been acknowledged as a representative metric in the data anonymization literature [8].”
Privacy: Theory meets Practice on the Map (2008, background): “Subsequent work [8] characterized the worst case knowledge needed to break the anonymity of a data-set and extended the work to releases of multiple views of the data [32].”
The cost of privacy: destruction of data-mining utility in anonymized data publishing (2008, background): “Other syntactic metrics include the number of generalization steps, average size of quasi-identifier equivalence classes, the sum of squares of class sizes [22], and preservation of marginals [16].”
Correlated network data publication via differential privacy (2013, methodology): “Given the degree frequency sequences of the original graph and the sanitized graph, F(G) and F(G̃), we measure their difference by Kullback–Leibler divergence (KLdivergence) [23]:”
Slicing: A New Approach to Privacy Preserving Data Publishing (2009, background): “It has been shown [1, 15, 35] that generalization for k - anonymity losses considerable amount of information, especially for high-dimensional data.”
Privacy in Search Logs (2009, methodology): “As notable differences, Brickell et al. [4] measure the utility with respect to data mining tasks and Kifer and Gehrke [11] develop specific techniques to boost utility with respect to log-linear models.”
A General Survey of Privacy-Preserving Data Mining Models and Algorithms (2008, methodology): “These include the applications of association rule mining [64], clustering [57, 62, 63] and collaborative filtering [104].”
M-invariance: towards privacy preserving re-publication of dynamic datasets (2007, background): “[15] A. Machanavajjhala, J. Gehrke, and D. Kifer. l-diversity: Privacy beyond k-anonymity.”
A Survey of Quantification of Privacy Preserving Data Mining Algorithms (2008, background): “Having this in mind, Kifer [17] proposes a utility measure related to Kullback-Leibler divergence.”
Worst-Case Background Knowledge for Privacy-Preserving Data Publishing (2007, background): “…extending our framework for probabilistic background knowledge, studying cost-based disclosure (since it was observed in [24] that not all disclosures are equally bad), and extending our results to other forms of anonymization, such as data-swapping and collections of anonymized marginals [21].”
Anatomy: simple and effective privacy preservation (2006, background): “RELATED WORK Generalization has been very well studied in the literature [1, 2, 4, 5, 6, 7, 8, 9, 10, 11, 12, 13, 14, 16, 17, 18].”
Minimality Attack in Privacy Preserving Data Publishing (2007, background): “An adversary may also have some additional knowledge about the individuals in the dataset or some knowledge about the data involved [12, 7, 13].”
On the tradeoff between privacy and utility in data publishing (2009, background): “the KL-divergence between the reconstructed distribution and the true distribution for all possible quasi-identifier values [13].”