Technology: Interaction data may allow identification of anonymized individuals over time
Papers suggesting that records of personal interactions can be used to identify a particular individual from anonymized datasets over time.Nature Communications. Will be published in.Our findings suggest that current practices in the handling of this type of data are likely to fail to meet the anonymization criteria set out in the European Union's General Data Protection Regulation.
Detailed data on personal interactions is collected by messaging apps, mobile carriers, social media providers and other apps and used for the operation and research purposes of those services.These data have been used to study individual interaction patterns, predict the spatial spread of epidemic diseases, and study the impact of friendships on political mobilization.Current data protection regulations allow data on personal interactions to be shared and sold without the consent of the user, provided that the data is anonymized.
Now, Yves-Alexandre de Montjoye, Ana-Maria Cretu and colleagues show that interpersonal interaction data is stable over time and can be used to identify specific individuals from anonymized datasets.The authors have developed a model using deep learning techniques and trained it to identify individuals based on a network of personal interactions, with a total of over 4 datasets collected over different periods of time. Applied to.This model was able to identify 2% of individuals based on a two-hop AC network (an AC network where individuals and subjects are two hops apart).The model also used an individual direct contact (2 hop) to identify people with a 52% chance.Due to the long-term stability of personal interaction, the books were able to identify 1% of people after 15 weeks using a two-hop exchange network.Furthermore, when this model was applied to a Bluetooth proximity dataset consisting of 2 people, there was a 20% or higher probability of individual identification.However, the authors say they do not believe that this model can be applied to contact tracing protocols (such as Google and Apple's contact notifications).
The results of this study show that the authors may be able to identify a particular individual over the long term from anonymized and disassociated data about interactions, which is important for compliance with privacy legislation. He claims to have it and states that security measures (access control systems, privacy enhancement systems, etc.) can be used to prevent re-identification.
[Original English »]
"Highlights of Nature Magazines" is a translation of a release made by Nature's public relations department for the press.If you need more accurate and detailed information, be sure to read the original paper.
* This article is reprinted from "Nature Japan Featured Highlights".
Reprinted from: "Technology: Data on interaction between individuals may allow long-term identification of a particular individual from anonymized data'
