Fraud has become a major issue for the insurance industry, with the Association of British Insurers (ABI) estimating that deceptive claims are responsible for adding an extra £50 to the premium of each customer’s annual bills.
The problem: this is hard to fight let alone spot in the first place. Insurance fraud attracts sophisticated criminal rings who are often very effective in circumventing fraud detection measures. In a typical hard fraud scenario, rings of fraudsters work together to stage fake accidents and claim soft tissue injuries. These fake accidents never happen, but are mere paper collisions complete with fake drivers, fake passengers, fake pedestrians and even fake witnesses.
Because soft tissue injuries are easy to falsify, difficult to validate, and expensive to treat, they are a favorite among fraudsters, who have even developed a term for them, “Cash for Crash”. Such rings normally include a number of roles. Fake collisions typically involve participation from professionals across several categories: doctors, who diagnose false injuries, lawyers, who file fraudulent claims, and bodyshops, which misrepresent damage to cars. Participants in the (false) accident will normally include drivers, passengers, pedestrians and witnesses.
Worse, fraudsters often create and manage rings by “recycling” participants so as to stage many accidents. Thus one accident may have a particular person playing the role of the driver; in another accident the same person may be a passenger or a pedestrian, and in another, the witness.
Clever use of roles can generate a large number of costly fake accidents, even with a small number of participants. In a six-person alleged collision you may have three false accident reports, with each person playing the role of “driver” once and “passenger” twice. Assuming an average claim of £20,000 per injured person, and £5,000 per car, the ring can claim an astonishing £390,000 in total. If ten people collude to commit insurance fraud, five false accidents are staged, and each person plays the role of the driver once, a witness once and a passenger three times, assuming an average claim of £40,000 per injured person and, again, £5000 per car, the ring can claim up to £1,600,000 for 40 people ‘injured’.
Clearly, we need a way to stop all this expensive criminality. Fortunately, while no fraud prevention measure is completely successful 100% of the time, detection methods designed to tackle the more intricate web of fraudulent identities and activity and help uncover proceedings before damage results – and making a huge contribution.
The methods in question are called graph databases. Graph databases specialise in identifying the relationships between multiple data points, which is how Google, for instance, has been able to exploit the connections in every Web document and build a connected dataset to rank search results, and how LinkedIn digitally harnesses real-life relationship networks via the data architecture of graphs so well.
So when considering insurance fraud detection, if we look at the data points and highlight their connections, we might be able to spot the clues to any potential fraud being carried out. The next frontier in motor insurance fraud detection, then, is to use social network analysis to uncover these rings. Connected analysis is perfectly capable of revealing relationships between people who are otherwise acting like perfect strangers – and stopping these nefarious rings building their schemes to ‘pay-day’.
How Graph Databases Can Help
Social network analytics tends not to be a strength of relational databases, the traditional business data system used in most insurance firms today. Discovering the ring requires a number of tables in a complex schema such as alleged/imaginary Accidents, Vehicles, Owners, Drivers, Passengers, Pedestrians, Witnesses, Providers, and joining these together multiple times — once per potential role — in order to uncover the full picture. Because such operations are so complex and costly, in computer performance, particularly for large data sets, this crucial form of analysis is often overlooked.
On the other hand, finding fraud rings with a graph database is a much simpler question (in technical terms) of following connections. Because graph databases are designed to query intricate connected networks, they can be used to identify fraud rings in a fairly straightforward fashion.
As a result, graph database queries can, and should, be added to the insurance company’s standard checks, at appropriate points in time— such as when the claim is filed—to flag suspected fraud rings in real time.
Two points are hopefully now clear. The first is the importance of detecting fraud as quickly as possible, so that criminals can be stopped before they have an opportunity to do too much damage. As business processes become faster and more automated, the time margins for detecting fraud are becoming narrower and narrower, increasing the call for real-time solutions like this even more.
The second is the value of connected analysis. Sophisticated criminals have learned to attack systems where they are weak. Traditional technologies, while still suitable, indeed necessary, for certain types of prevention, are not designed to detect elaborate fraud rings, so we need to look to graph databases to add value.
Frauds that were previously hidden become obvious when looking at them with a system designed to manage connected data, using real-time graph queries as a powerful tool for detecting a variety of highly-impactful fraud scenarios.
And we need to have these out there protecting our money and keeping our premiums low.
By Emil Eifrem, CEO, Neo Technology.