-
Banks are using technology developed by search engines to cheaply sort through vast amounts of consumer data to identify potential fraud.
September 1
Zions Bancorp., in its attempt to spot fraud, was gathering much more data than its technology could handle. While searching for a needle in a haystack, it was just piling on more hay.
To finally make sense of the data it was collecting, it had to replace its process with a faster and less expensive link analysis tool.
"Many years ago we started aggregating security event information into a relational database to mine the data, correlate it and turn the data into more useful intelligence … [but] the relational databases have scale issues, and at a certain point you are not able to add more horsepower to get what you want out of it," says Preston Wood, the Salt Lake City company's chief information security officer.
After many years of storing this data, the process of retrieving it had become too slow.
To speed things up under its old system, Zions would have had to add to the hodgepodge of products it was using from Microsoft Corp. and Oracle Inc.
Instead, early this year, Zions turned to Zettaset Inc. of Mountain View, Calif. The vendor is a specialist in link analysis, a technology in use for a decade, which it has updated using search engine techniques. Banks have only recently started to use such systems.
"Now I have one repository to get linkage and gain information, whereas in the past it may have been sitting in a bunch of different locations and difficult to link up," Wood says.
Link analysis can help banks like Zions examine structured and unstructured data. Structured data is often the kind found in form fields. Unstructured data, as its name implies, is more random. It is often text-based and does not fit well into relational databases, experts say.
Link analysis has important fraud-fighting potential, since fraudsters leave a trail of data that spans bank silos and channels. A good link analysis tool can look at data both inside and outside the bank.
Large relational databases capable of handling similar loads of information could cost millions of dollars, experts say. Zettaset bases its technology on open-source code called Hadoop, which lowers the price. Systems like the one it helped Zions to install cost $500,000 or less, it says.
"In less than a week we had a massive, multinode cluster set up on [Zions'] site with our software installed," says Brian Christian, Zettaset's chief executive.
Now Zions can identify security threats more quickly, allowing it to become more proactive when it sees abnormal behavior. It can also analyze new and evolving security threats against the bank, Wood says.
"Once you have been able to pull customer data and transaction data and you have a detailed level of information across the environment, the sky is the limit on what intelligence you can derive out of that," Wood says.
Systems based on open-source code, such as what Zions uses, have not been tested for very long in a dedicated banking environment, says Julie Conroy McNelley, senior risk and fraud analyst at Aite Group LLC.
"The solutions that have been in the market for some time have an advantage, because they have the benefit of the years of refinement of their analytic approach to help minimize the noise and maximize the value," McNelley said by email.
At the same time, banks' data needs are evolving, experts say.
"Bankers will be able to analyze petabytes of identity, transaction, social and other data, [and] may not need to integrate, cleanse and store it themselves," says Michael Versace, research director for global risk services at IDC Financial Insights.
