Modeling and Evaluation of k-anonymization Metrics - Archive ouverte HAL Access content directly
Conference Papers Year : 2020

Modeling and Evaluation of k-anonymization Metrics


The k-anonymity concept introduced in (Samarati and Sweeney 1998) proposes a good trade-off between the privacy and the utility of the data published for exploitation. However, minimizing the loss of information throughout the k-anonymization of a database is known to be NP-Hard (Meyerson and Williams 2004). Several previous works defined metrics to measure and to optimize this process according to different priorities or different ways of looking at things. In this paper, we first present a unified modeling of the optimization metrics for the k-anonymization of a database. Then, we propose different new metrics for this optimization problem. Finally, we evaluate three metrics of the literature and our new metrics using a greedy algorithm along the anonymization process for 21 values of k.
Fichier principal
Vignette du fichier
paper_18.pdf (1.04 Mo) Télécharger le fichier
Origin : Files produced by the author(s)

Dates and versions

hal-03236397 , version 1 (26-05-2021)


  • HAL Id : hal-03236397 , version 1


Clémence Mauger, Gaël Le Mahec, Gilles Dequen. Modeling and Evaluation of k-anonymization Metrics. Privacy Preserving Artificial Intelligence Workshop of AAAI 2020, Feb 2020, New York (virtual), United States. ⟨hal-03236397⟩
127 View
513 Download


Gmail Facebook X LinkedIn More