Follow Datanami:
July 18, 2013

Fighting Crime with Big Data

Alex Woodie

In today’s complex world, white collar criminals often hide their nefarious deeds behind a cloak of obfuscating data. Now prosecutors are beginning to turn the tables with high powered analysis tools that allow them to not only identify crimes and criminals from patterns in the data, but to present their cases in court in a compelling way.

Successfully prosecuting white collar crimes, such as insider trading, money laundering, and healthcare fraud, has traditionally been a very difficult task, says Ryan Taylor, a former litigator who now works as legal counsel for Palantir, a company that makes big data analysis and presentation tool for prosecutors.

“The historical approach is, ‘Let’s through a lot of people at the problem, let’s have them the review documents in a linear fashion. If they’re not doing it quickly enough, let’s get more people,'” Taylor says in a recent video. “Everyone who is reviewing documents in a linear fashion is looking for specific nuggets in individual documents. But there’s no way if you take that approach, to be able to draw a connection between all the documents in the case.”

Palantir’s platform, called Gotham, uses big data analysis techniques to enable investigators to bring together large amounts of structured and unstructured data, and to analyze it for potential evidence of a crime. The software also uses visualization tools to help prosecutors present the analysis to a judge, jury, or fact-finder in a compelling manner.

The idea is not about finding one or two documents that are “smoking guns,” Taylor says. “Instead, you have to paint a compelling story that shows the circumstantial evidence, how it plays together, and how the sequence of events over time and space demonstrates the evidence of the crime,” he says. “Being able to bring that data together and actually build up the sequence of events is something that prosecutors can do much more effectively in Palantir than they could before.”

Money laundering can be one of the most complicated cases to bring to prosecution, Taylor says, “in part because the whole goal of money laundering is to obfuscate the flow of transactions and to make it as confusing as possible. Often times prosecutors have historically chosen to purse the other crimes [associated with money laundering] instead of the money laundering crime, because money laundering can be so complicated and difficult to prove.”

Palantir Gotham is changing that with its capability to track the complex flow of money across multiple banks and accounts–specifically its capability to navigate through the various formats used by banks. By culling data from different sources and to present it as a single ontology that connects illicit actors conspiring in a coherent scheme, prosecutors have been able to get convictions without spending thousands of man-hours manually constructing the events.

Healthcare fraud is similar to money laundering in some ways, in that it can involve a wide range of players. But it adds an element of complexity by requiring a prosecutor to successfully bring together data residing in many different formats and databases, such as medical claims, tip lines, OIG Exclusion Lists, and others.

“Before Palantir, there was really no effective infrastructure for …. collaborating across those different platforms,” Taylor says. “A lot of the different information sources were very specific, and required unique log-ins and was silo’ed in a way that you couldn’t connect the dots between them.”

An individual claim may be fraudulent, but you wouldn’t know that by looking just at that claim. “The people committing the fraud on those claims aren’t going to make that transaction look suspicious,” Taylor says. “It’s a normal prescription, or a straight forward claim. But when look at it in conjunction with the rest of the data, you can paint a much more compelling story to be able to present at prosecution.”

For example, by connecting known aliases of crooked doctors from the OIG Exclusion List with geographic data culled from Google Maps, an investigator using Palantir Gotham could see that a pharmacy located in an undeveloped lot is possibly a phantom pharmacy that isn’t filling any prescriptions and is likely committing fraud.

Palantir works with a number of prosecutors, including the federal Department of Justice, several U.S. Attorneys offices, state attorneys general, and various local district attorney offices, Taylor says. In addition to the crimes mentioned here, the software has been used with cases involving consumer fraud, public corruption, drug and human trafficking, and gang violence.

Related Items:

Finding Value in Data Through Life Saving Applications 

IBM: Five Use Cases Driving Big Data Adoption 

New York City Calls in the Code Corps