Talk:K-anonymity

Randomization included?
it appears that Kishor has changed gender from Male to Female in the anonymization step. Is the article intended to suggest that k-anonymization also permits explicit falsification of data in order to acheive some k-anonymous property? — Preceding unsigned comment added by Dkgdkg (talk • contribs) 01:50, 22 April 2014 (UTC)


 * No, that would be in error and is no longer the case. WilliamKF (talk) 00:49, 22 May 2014 (UTC)

First introduced 1998 by Samarati/Sweeney?
The introduction states, that the concept of k-anonymity was first introduced in 2002.

However, the paper "Protecting privacy when disclosing information: k-anonymity and its enforcement through generalization and suppression", released in 1998 by Samarati/Sweeney clearly introduces those concepts a few years earlier.

Is my understanding off, or is the information in the article just incomplete? 42ndTowel (talk) 23:16, 28 April 2015 (UTC)


 * The information in the article is complete in the sense that it formally develops the definition of k-anonymity. The issue is with the date of publication.  A search of IEEE Explore, ACM Digital Library, and DBLP revealed no refereed papers pre-dating the 2002 journal paper. A formal definition based on the journal article will vastly improve the quality of this article. Vonkje (talk) 14:10, 23 July 2015 (UTC)


 * I believe the issue is that the earlier paper was a tech report published by SRI . Not being peer reviewed or published by a publishing house, the IEEE, ACM, and DBLP probably don't consider it significant enough to list.  However, I believe it would count as a publication for legal purposes such as patents.  Thus, I suggest taking the earlier paper as the first.  Duckpigcow (talk) 05:54, 14 October 2016 (UTC)


 * There also appears to be an abstract from 1998, maybe based on the tech report, in ACM's library . Duckpigcow (talk) 06:02, 14 October 2016 (UTC)

clarify paragraph
This makes no sense to a layman; there is no context and the terminology is not covered by this article:

For example, researchers showed that, given 4 points, the unicity of mobile phone datasets ( E 4 {\displaystyle {\mathcal {E}}_{4}}, k-anonymity when k = 1 {\displaystyle k=1} ) can be as high as 95%. — Preceding unsigned comment added by 100.1.206.144 (talk) 01:44, 9 July 2018 (UTC)

Password hashing is not k-anonymity
Despite the accurate descriptions of "media coverage", the description of breached password checking as using k-anonymity seems misleading. k-anonymity is meant to keep a database private, but the mechanisms for password checking does not keep the database private at all. It only protects the client's privacy. This should be mentioned in the article. — Preceding unsigned comment added by J2kun (talk • contribs) 16:32, 4 June 2022 (UTC)

Significantly expanding the example
I intend to significantly expand the example so that it is more realistic and so that it shows some of the problems with k-anonyminity. I intend to do this by adding three more columns: weight, blood pressure, and height. The idea is that under anonymity these are seen as "non-identifying" values. However, I'll show that they can be identifying, either because they are extremes, or because there can be another document that links with the values. I'll provide citations. Simsong (talk) 23:47, 24 December 2022 (UTC)

https://de.wikipedia.org/wiki/Anonymisierung
https://de.wikipedia.org/wiki/Anonymisierung 2001:1260:238:89E0:1C83:4628:7BD4:8C81 (talk) 18:36, 12 November 2023 (UTC)