Financial fraud is an increasingly pervasive problem, estimated to cost more than $5 trillion in 2019 – more than 6% of the Global Domestic Product. Stemming the wave of financial loss requires constant vigilance since fraud perpetrators continue to evolve their tactics, allowing them to evade detection.
Nik Vora is the Vice President – APAC, at leading Graph Database Platform, Neo4j. Prior to this, he established Qubole in APAC and had the opportunity to lead and be a part of 3 hyper-growth GTM plays for Capillary Technology, a Cloud-SaaS platform. Over the last 10 years, his focus has been about building markets in Singapore, Australia, India and China. Here he explains how traditional fraud detection methods often fail to minimize losses since they perform discrete analyses that are susceptible to false positives and negatives. Knowing this, increasingly sophisticated fraudsters develop a variety of ways to exploit the weaknesses of discrete analysis.
Winning the battle against money laundering requires a technology that better harvests information from transactions – and other sources – and better detect suspicious activity in real-time and at scale. This has been challenging because companies process billions of transactions per day involving tens of millions of parties.
The steps below outline a typical graph approach to fraud detection.
1. Create a graph of relationships of information about individuals. Connect all available information: account IDs, user names, account numbers, names, IP addresses, social media accounts, email addresses, identification numbers, mailing addresses, dates of birth, and so on.
2. Define what suspicious activity to look for. For example, consider:
Common attributes (same email addresses, tax identification number, or phone number, for instance)
Multiple parties using the same account
Short paths between transactions (a rapid return of purchase with no support call or reason given, for example)
3. Run graph queries on these attributes or use graph algorithms to investigate. Algorithms surface fraud rings by detecting discrete islands of activity or groups that interact more with each other than the rest of the graph or network.
4. Explore result sets with a visualization tool to investigate and verify fraudulent activity. Graph visualization cuts the manual analyst review time in half, allowing them to stop fraudulent transactions sooner and reduce wait times for non-fraudulent customers.
5. Automate the steps above by converting graph algorithm scores into features to add to your machine learning model so that you identify fraud faster, minimize false positives, and shut it down sooner.
Graph technology is the ideal enabler for efficient and manageable fraud detection solutions. From fraud rings and collusive groups to educated criminals operating on their own, graph database technology uncovers various important fraud patterns – and all in real-time.