big data
We continue our Big Data Week on Kaspersky Business. The first two posts of the series are available via the links below:
World of finance loses big on fraud of all kinds – by some estimates they lose up to dozens, or even hundreds of billion of dollars annually on fraudulent transactions alone. No surprise that it wanted to do something about this. First antifraud measures were implemented by Visa, for instance, decades ago, but up until rather recently, Visa has acknowledged its antifraud analytic potence was limited: older analytic engines could only study 40 aspects of a transaction at once, attempting to detect patterns associated with frauds. Earlier analytic models studied as little as 2% of transaction data. Given that a typical statistical error is 3-4%, it doesn’t look good enough.
A bigger stick for fighting bigger fraud. Is victory there? – Nope
Tweet
So Visa turned to Big Data – to the systems capable of finely grinding all those huge amounts of data – namely, Apache Hadoop, derived basically from Google’s technologies such as MapReduce. Currently its analytic engine is capable of studying as many as 500 aspects of a transaction at once, and uses as many as 16 models of possible fraud patterns at once.
According to WSJ’s Steve Rosenbush, the company estimates that the model has identified $2 billion in potential annual incremental fraud opportunities, and gave it the chance to address those vulnerabilities before that money was lost.
Does it mean, though, that the fraud is beaten? There are lots of publications and proclamations on how Big Data revolutionizes fraud prevention, but none speak of total victory. One can combat fraud with more powerful weapons, but it’s not eradicated, and probably won’t be.
Fraudsters read the same books as Big Data white-hat experts.
Tweet
Experts acknowledge that the cyberfraudsters aren’t standing still either: As John Kunze, CEO of a digital money transfer company Xoom Corporation puts it, “The fraudsters are reading all the books we are. They are PhDs themselves.” And if these bad guys see the big protective wall built around the payment services from the Big Data grinding material, they are looking for the ways to climb over or walk around wherever possible.
And they are clever enough to succeed at least occasionally.
Big Data Week
Tips