Applying Cognitive, Big Data Analytics to Thwart Fraud

Share this post:

A few weeks ago, in Washington, D.C., I had the privilege of moderating a MeriTalk panel, “I Spy Fraud – Before it Happens,” where we delved into the ongoing battle against fraud. But just prior to the discussion,  US Government Accountability Office Assistant Director Linda Miller, said something interesting. She said, the traditional approach to fighting fraud, known as “pay and chase,” is shifting to a focus on stopping it before it happens. I couldn’t agree more.

But I would add that the new approach requires advancing analytics to become more predictive, cognitive and innovative.

The biggest challenge is quite simply the enormity of the data. As we all know, the federal government is one of the biggest users – and generators – of data. US Geological Survey and NASA data exceeds 7.5 petabytes (equal to 100 years of HD video). How can agencies make sense of all that data and use it to get insights? And how usable is the data after it ages? And how does new data change the meaning of the old data?

Our approach with federal agencies is to help them take this vast amount of information from all different sources and be able to develop knowledge and understanding.

Agencies understand analytics and I believe they are firmly ready to take things further with a cognitive approach. In IBM’s just released Institute for Business Value report, Mission Possible: Your Cognitive

Future in Government, we have some interesting stats:

• 87% of government officials familiar with cognitive computing believe it will play a disruptive role in their organization.
• 83% of government officials familiar with cognitive computing believe it will have a critical impact on the future of their organization.
• 100% of government officials familiar with cognitive computing intend to invest in cognitive capabilities.

What does this mean for fraud? Government leaders must be smarter in how they approach data. While the digital age has provided governments with a massive amount of data brimming with insights, organizations still struggle to unlock its full value. Advances in the pioneering area of cognitive computing can help bridge the gap between data quantity and data insights. It can also help deal with ambiguity of data, which Johan Bos-Beijer, Director of Analytics Services from the General Services Administration, stated during the panel has helped enable more fraud.

Cognitive-based systems can build knowledge, understand natural language and provide confidence-weighted responses. And these systems can quickly find the proverbial needle in a haystack, identifying new patterns and insights – something particularly relevant in complex government information environments.

Another area “sparking” interest among federal agencies is the use of open source, including Spark. As a platform, it offers several major advantages that align with the future of analytics and the needs of enterprises — it is open, precise, fast, and supports agile data science and development. And it will better enable developers to create “smart apps” that are driven by insights and intelligence much faster.

With Spark, a data scientist can look across a vast corpus of data, start to understand the patterns and run algorithms or machine learning on top of that to really change how you can gain analytical insights. For those agencies focused on preventing fraud, being able to meld together streams of information, coming from many different sources and often in real time, can create an opportunity to decrease improper tax filings, prevent crime and reduce fraud – for example.

When it comes to fraud, I believe we now have the capabilities at our disposal to get away from “pay and chase” approaches and help stop fraud before it happens.
______________________________________

For more about Andrew’s perspectives on fraud, listen to a podcast interview with him by MeriTalk Executive Editor Dan Verton.