L-diversity: privacy beyond k-anonymity

By Ashwin Machanavajjhala, Johannes Gehrke, Daniel Kifer, Muthuramakrishnan Venkitasubramaniam. 2006

5,621 citations recorded by the source corpus.

Subfields: Artificial Intelligence, Privacy & Security

How researchers see this paper

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

In Artificial Intelligence

In Privacy & Security

Citations in the corpus

Privacy in Deep Learning: A Survey (2020, background): “This has led to privacy notions such as l-diversity [52] and t-closeness [53], which are out of the scope of this survey.”
Prochlo: Strong Privacy for Analytics in the Crowd (2017, methodology): “iduals [65,66]. However, since k-anonymity does not limit statistical inferences drawn indirectly, from subgroup intersections or differences, it cannot offer mathematically-strong privacy guarantees [32,48]. In the last decade, the deﬁnitions of differential privacy have become the accepted standar”
Privacy Preserving Data Mining (2015, methodology): “Such techniques are: randomization method (Agrawal and Srikant, 2000; Agrawal and Aggarwal, 2002), k-anonymity model and l-diversity (Machanavajjhala et al ., 2007), distributed privacy preservation and downgrading application effectiveness. k-anonymity captures protection of released data against…”
On Safeguarding Privacy and Security in the Framework of Federated Learning (2019, methodology): “These generic approaches include de-identification methods like anonymization [5], obfuscation methods like differential privacy [6], cryptographic techniques like homomorphic encryption [7] and secure multi-party computation (SMC) protocols like oblivious transfer and garbled circuits [8].”
Differentially Private Password Frequency Lists (2016, background): “In response to de-anonymization attacks researchers began to propose formal data privacy models like k-anonymity [35], [32] and `-diversity [25].”
Dependence Makes You Vulnberable: Differential Privacy Under Dependent Tuples (2016, background): “Data privacy is an issue of critical importance, motivating perturbation of query results over sensitive datasets for protecting users’ privacy [5], [12], [26], [29], [31], [38].”
Data Synthesis based on Generative Adversarial Networks (2018, background): “To protect against these attacks, the authors of [24] introduced l-diversity to ensure that the sensitive attributes”
Differentially Private Empirical Risk Minimization (2009, neutral): “There has also been a significant amount of work on privacy-preserving data mining (Agrawal and Srikant, 2000; Evfimievski et al., 2003; Sweeney, 2002; Machanavajjhala et al., 2006), spanning several communities, that uses privacy models other than differential privacy.”
A Statistical Framework for Differential Privacy (2008, neutral): “Disclosure limitation methods include clustering (Sweeney 2002; Aggarwal et al. 2006), -diversity (Machanavajjhala et al. 2006), t-closeness (Li, Li, and Venkatasubramanian 2007), data swapping (Fienberg and McIntyre 2004), matrix masking (Ting, Fienberg, and Trottini 2008), cryptographic…”
Composition attacks and auxiliary information in data privacy (2008, methodology): “However, as pointed out in [26], the actual anonymity offered is less than this ideal value and is equal to number of distinct values in each equivalence class.”
Differentially private aggregation of distributed time-series with transformation and encryption (2010, background): “However, [20] pointed out that if the time-series exhibit certain patterns, then independent perturbation of query answers can be distinguished from the original answers and filtered out.”
Achieving anonymity via clustering (2006, methodology): “Extensions to the k-Anonymity model, like the notion of l-diversity [Machanavajjhala et al. 2006], can be applied independently to our clustering formulation.”
PrivateClean: Data Cleaning and Differential Privacy (2016, methodology): “These techniques include value aggregation, value suppression, and homeomorphic encryption [26,28,38,42].”
Anti-discrimination Analysis Using Privacy Attack Strategies (2014, methodology): “Note that in each group of our example table, every sensitive value is distinct and so the group size is equal to the parameter l in the l-diversity privacy model [11].”