Fraud Detection and Prevention – From Post-It note to production part 3/3
In the final post of this series we will be looking briefly at what the future holds for the fraud detection and prevention system on the ServiceNow platform. The core system has been running now for almost three months and the Virgin Trains compliance team have seen some great results, now we need to turn great into amazing.
The first planned major addition to the system is the inclusion of machine learning. As it stands right now, the system relies upon human built rules which are used to detect elements of fraud, these can be tweaked by hand to better fit the current climate, however plans are being put together to hand off this ‘tweaking’ the ServiceNow platform itself. Using the results of previously raised fraud cases, the platform will assess the current effectiveness of the rules involved and identify areas for improvement. Of course, machine learning is not perfect, so it is planned that the platform will suggest tweaks that need to be made and send these for approval by a human being before being implemented. As far as we are aware, this use case is a first of its kind for the ServiceNow platform, which is an exciting position for both UP3 and Virgin Trains to be in.
As discussed in the previous post, digital image forensics is likely to play a part in a new phase of the system. The plan is to implement a client-side tool that can be used by compliance agents to detect modified images, more specifically, areas of an image that have been edited. This can be achieved using various filters over an image, one such filter is error level analysis. As you can see in the image below, you can use this filter to highlight areas of the image where edits have likely taken place.
Whilst these techniques are not 100% accurate, they can often help to confirm already suspected edited images, such as those images that have already been detected by the system as being edited by unauthorised software.
The last addition we can discuss about the next phase is the creation of a fraud case export tool. The idea behind this is to reduce the amount of manual work needed to create the documentation required to hand a case over to the British Transport Police. By using data already in the system, we can easily create a full data export in PDF / CSV format including full information on all of the consumers journeys, meeting notes, consumer details and their connections. The Virgin Trains Compliance team have already estimated that the addition of this functionality is likely to save them hours of work for each case, which can now be spent further investigating other potentially fraudulent activity.
At UP3 we are very excited about the next phase of this project. We will be sure to provide an update upon completion, showing you all the new functionality, including things not discussed here.
If you have any questions, or you’re interested in a similar or other solutions, please contact us, it would be great to hear from you.