Neo Technology’s Emil Eifrem looks at how graph technology can help governments, companies and professional fraud investigators piece together evidence against money laundering operations
Money laundering happens across the globe, it underpins criminal and terrorist activities and it is on the increase leaving organisations, governments and law enforcement agencies firefighting.
Recent US government estimates show that money laundering sits at around US$800 billion to $2 trillion. That is 2% to 5% of the global GDP. But it isn’t just banks and financial institutions that are being hit. It is any organisation that carries out financial transactions – from retailers to insurance companies.
Governments are increasingly looking for ways to combat money laundering. The European Union, for example, has recently strengthened legislation to tackle terrorism financing, tax avoidance and money laundering in Europe. But, despite tightened legal measures and the Panama Papers and Russian Laundromat exposures, money laundering shows no signs of abating.
The problem is that these financial crimes are broad and complex and can take many forms. The perpetrators, often part of a large network, hide their tracks carefully to avoid large fines and prison sentences.
That means creating endless ‘millefeuille’ layers of fake identities, middlemen and multifarious financial schemes, making it extremely difficult and time consuming for law enforcement agencies, financial institutions and fraud detection professionals to correlate evidence against them.
Sifting through big data to find any odd or conflicting patterns that may lead to recognising a money laundering transaction is an enormous task. Relational databases are not flexible enough to tackle changing, complex data structures. NoSQL databases don’t underscore relationships as primary citizens. In both cases, relationships in the data are either missing or hard to compute.
A traditional relational database stores data in separate tables and columns, requiring complex code to join up relationships. A graph database, however, stores the data as “nodes” or relationships, which can be easily linked.
Graph technologies are powerful enough to join the dots and uncover discrepancies. Take for example the Panama Papers. The International Consortium of Investigative Journalists (ICIJ) had more than 200 journalists in 64 countries working on cross-border investigations. This required the processing of 2.6 terabytes of data and mining 11.5 million files. The ICIJ used graph technology to uncover the connections because it was able to make the link between relationships. In addition, journalists did not have to be data scientists to work with a graph representation of a dataset, but could simply click their way through the data while following connections.
Graph technology can be used to deliver a holistic view of the various entities involved in financial crime and the relationships between these entities, which nine times out of ten expose the hidden, fraudulent connections.
This means accounts that seem separate and unrelated can be discovered as actually linked – like having the same IP addresses for an account, or with different account names that turn out to be one and the same. This allows money risk management teams to much more easily dig out potential money laundering rings in the form of graph patterns.
They also empower users to do a lot more. Using a graph database, for example, we can start asking questions like, ‘Are my customers in contact with members of criminal networks?’ or ‘Am I doing business with companies with ties to countries on a sanctions list (e.g. the Specially Designated Nationals List (SDN)?’ That’s very straightforward with a graph database, as it can sweep up attributes on individuals to form the bigger picture.
The Neo4j graph database also provides the ability to quickly store and query large, complex and evolving datasets. That’s important, as a financial institution needs information about the companies and individuals it does business with and their transactions, obviously – but with graphs that core data can be augmented with other data sources like watch lists, company registries, etc. A graph approach can help consolidate these different entities coming from various systems into a single view, and as the data is updated with new transactions, persons, companies and relationships, the graph database can be used as a magnifying glass to help spot (potentially) suspicious connections.
To sum up, graphs helps investigators discover new, hidden connections. Increasingly, it is not enough to identify individual suspicious individuals, but the complicated networks in which they operate. Traditional technologies, whilst still having a place in fraud prevention, are not powerful enough to encompass the full network of criminality and thus to significantly aid financial institutions fighting fraud, anti-money laundering or terrorism funding.
Graph database technology therefore has the power to unearth data and connections that can lead investigators directly to money laundering cases from what may initially seem like totally unconnected events. This innate ability is opening up a whole new way of detecting money laundering as it is happening, via real time analysis of data relationships.
At long last, have we finally found a way of staying one step ahead of the bad guys? The evidence looks pretty convincing to our customers.
The author is CEO of Neo Technology, the company behind the world’s leading graph database Neo4j (http://neo4j.com/)