Welcome Stranger to OCC!Login | Register

Finding People in a Credit Card Metadata Haystack

Category: Science & Technology
Posted: 08:16AM

Many of us would probably like to think that we can be hidden in a crowd and that without personal information, identifying an individual from a massive database is going to be very difficult. Well, it looks like that is not the case as researchers at MIT have found that they can accurately identify people out of datasets of over one million people, using just three, imprecise, pieces of purchase information.

For this work, the researchers worked with a data set that contained the names and locations of shops, days purchases were made, and the amounts of each purchase. Purchases made with the same credit card were given the same identification number. The researchers then analyzed the data set for any similar purchase patterns, between the identification numbers. They found that with just two data points, without price information, they could identify 40% of the people in the data set of over one million people, and with five points, almost everyone could be identified. They then tried coarsening the data, such as considering ranges of purchase amounts the same and purchases made the same week instead of the same day, and found that just four purchases were needed to identify 70% of the people. This coarsening was in part done to emulate if someone had to guess at the information, like they had looked at a shared pictured and not multiple receipts.

Obviously this shows that in Big Data, it is still possible to find specific people, without specific information. The researchers, as part of separate work, are also working on a system that would give people more control over their information, limiting what third-parties can access to only pertinent data.

Source: MIT

Register as a member to subscribe comments.

This news has comment postings disabled because it is now archived.

© 2001-2018 Overclockers Club ® Privacy Policy
Elapsed: 0.0891499519   (xlweb1)