No chance for insurance fraud

No chance for insurance fraud

With a Big Data solution from GFT, insurance companies can now assess whether someone is trying to pull the wool over their eyes as soon as a claim is reported.

Hanover/Stuttgart, 11 March 2014 - Insurance fraudsters are becoming more and more professional. In many cases, the insurance companies have absolutely no way of detecting fraudulent claims. This is because they lack the database they need to identify patterns or establish striking connections to other insurance claims. GFT's solution, based on Apache Hadoop and other Open Source components, brings all of the required data together in an inexpensive product.

A short pause at the junction, a deceptive wave to the other car to allow them to pass, foot on the pedal – and crash. The other party doesn't stand a chance, so all that remains is for him to report it to his insurance company. There are thousands of crashes every day on Europe's roads, and more and more of them are deliberate. In most cases, the insurance companies pay out because it is only rare that the circumstances of the crash are suspicious. Their costs rise, and as a result so do the premiums charged to customers.

"In the past, insurance companies were virtually powerless against professional insurance fraudsters," says Marika Lulay, Member of the Board of the GFT Group and the person responsible for GFT's operations. "Conventional internal IT solutions designed to uncover fraud are too simple and haven't evolved to take account of the wiles of criminals. Insurance companies were put off by the costs and amount of work involved with the complicated solutions from third parties that may have been able to help. Now, because we're able to analyse large and heterogeneous volumes of data virtually in real time with new Big Data solutions, the tide is turning."

Insurance companies can now identify suspicious similarities to previous damage claims as soon as the claim is submitted. For example whether the third party has already been involved as the damaged party in the past. Or whether the location at which the accident happened has similar claim patterns. If so, further investigatory steps can be taken immediately after the claim is made.

At CeBIT, GFT is unveiling a Big Data solution that can cut the infrastructure costs required to a minimum, thanks to various Open Source components. The experts at GFT have developed a scalable solution based on the Apache Hadoop platform that can scrutinise over 60 million data records for suspicious similarities in seconds. The Open Source search engine ElasticSearch makes real-time enquiries possible, while OrientDB, an Open Source database with a graphical interface, displays the relationships that are uncovered in the morass of data in an easy-to-understand format.

"Of course, our experts also ensure that the various databases that insurance companies typically have are standardised. This ensures that different data formats from existing database systems are merged," says Marika Lulay. "Now, the companies are finally able to mine the riches of their data."

"Conventional internal IT solutions designed to uncover fraud are too simple and haven't evolved to take account of the wiles of criminals. Insurance companies were put off by the costs and amount of work involved with the complicated solutions from third parties that may have been able to help. Now, because we're able to analyse large and heterogeneous volumes of data virtually in real time with new Big Data solutions, the tide is turning." ”
Marika Lulay Chief Operating Officer at GFT
Quote

Video: Insurance fraud detection real time scoring