8 ms·
General question to HN because i don't know the answer. Has anyone ever tried de-duplicating and matching up records between multiple leaks to build user/perso
by swarnie_ 7y ago
General question to HN because i don't know the answer.
Has anyone ever tried de-duplicating and matching up records between multiple leaks to build user/person profiles?
I guess with enough unique keys, compute power and time you could build a reasonably accurate profile of a person by matching up email addresses, phone number or SN numbers?
Would probably make identify thief and fraud a lot easier going forwards.
- dannyw 7y agoYou have, quite literally, described Facebook.
- anon73044 7y agoProbably not much more than what was in the Equifax breach. I think one of the bigger issues in generating profiles on ~325 million people would be merging tons of incomplete data and lots of old/outdated data. Not to say that you couldn't use some machine learning to fill in the blanks, but I'm sure most if the big data/analytics companies are already doing that.
- astura 7y agoData broker-type companies do this already and sell it. They even include stuff like your probable religion, probable income/net worth, probable hobbies/interests, etc. For example: >Each record contains entries that go far beyond contact information and public records to include more than 400 variables on a vast range of specific characteristics: whether the person smokes, their religion, whether they have dogs or cats, and interests as varied as scuba diving and plus-size apparel. https://www.wired.com/story/exactis-database-leak-340-million-records/ https://www.wired.com/story/exactis-database-leak-340-millio... HN Discussion: https://news.ycombinator.com/item?id=17421140 https://news.ycombinator.com/item?id=17421140