Fighting Fraud and Money Laundering With Graph Data Science

The impact of money laundering permeates both the economic and social spheres worldwide. With the power to collapse banks, money laundering has placed an enormous responsibility on financial institutions to ensure that financial crimes are appropriately detected and thwarted. By failing to do so, financial institutions and other businesses alike face possible fines, legal implications, and the loss of reputation.

While banks have been on the lookout for financial crimes with Anti-Money Laundering (AML) laws in place, as digitalisation occurs and criminals continuously shift to the following scheme, old AML tools simply can’t keep up. With the sophistication of the modern criminal that is always on the lookout for new exploits, the financial industry is faced with an insurmountable amount of pressure as money laundering and fraud cases increase and the task to find new criminal activities becomes much more complex.

According to PwC’s “Global Economic Crime and Fraud Survey 2020,” 47% of companies had experienced fraud. Along with the bigger umbrella of money laundering, inspectors have difficulty tracking these trails, having to depend on measly crumbs of information and having to pore over reams of data. It’s incredibly inefficient and time-consuming to catch the bad guys using conventional database technologies.

Although data scientists have developed ML models to detect fraud, they have overlooked the importance of the network structure. During investigations, when looking at singular accounts, it is easy for investigators to miss essential connections without the proper visualisation tools. Ordinarily, investigators looking at relationships and commonalities between the data may have more success when catching the bad guys.

Tabular data models just aren’t equipped to handle the complexity of the relationships that authorities require, as join tables and queries take up a significant amount of time and effort. What authorities need in these cases is a database built on connections as opposed to collection. Neo4j Graph Data Science allows exploration and analysis through searches, queries, and graph algorithms. With these features, data scientists can find patterns predictive of fraud to add them to their machine-learning models. This increases the predictive accuracy of fraud detection strategies.

As fraud is a high-cost crime, Neo4j has been used successfully by many financial institutions. One major fraud and money laundering case that used Neo4j was the International Consortium of Investigative Journalists (ICIJ) Pandora Papers investigation.

The Pandora Papers investigation was based on the data leak of 14 off-shore service providers. The leak contained 2.94 terabytes of data, which included files that held the unsavoury financial dealings of politicians, billionaires, celebrities, fraudsters, drug dealers, and royals world-wide. With the help of Neo4j’s graph platform, the journalist collaboration was able to explore connections within all that data. Exposing many financial crimes, including fraud, money laundering and tax evasion.

Another example of Neo4j’s success has been with a financial services client that needed to speed up their review process. This institution needed to detect fraud rings that are sophisticated in a relatively short amount of time. With the application of Neo4j, the financial institution was able to cut their manual review time from eight to three minutes. The decision to employ Neo4j also revealed fraud activity that had gone unnoticed.

Head on to their website to learn more about how Neo4j can help detect and stop fraud and money laundering activities.

share us your thought

0 Comment Log in or register to post comments