Print Save PDF

About 5 minutes

Watch any con game movie – Ocean’s Eleven, The Sting, Catch Me If You Can, and Dirty Rotten Scoundrels – and what do you see?

There is always a team of people (never less than 2 or 3 individuals) involved in pulling off a successful con. It’s no different in the world of online fraud.

Yes, some schmuck with a Hotmail account will try to phish you by pretending he represents Bank of America’s fraud protection team.

But even that guy rarely acts alone. He usually has to have at least one other sidekick in the background if he ever hopes to elicit account details, withdraw funds from the unsuspecting dupe, and avoid detection.

More often than not, sophisticated crime rings sit in the back of these scams.

There are even tales of vast office complexes in parts of Eastern Europe dedicated to phishing, ransomware, and other fraudulent practices.


The Real Business Costs of Cyber Threats and Fraud

These types of cyber threats are not small, isolated issues. One company lost $47 million when someone posed as its CEO and coaxed funds out of the finance department for a supposed acquisition.

The FBI notes tens of millions being collected in Bitcoin by ransomware specialists and a recent survey from Datto suggests that businesses have paid out $375 million over the last year resulting from ransomware attacks.

And then there are a legion of fraudsters applying for credit cards, loans, overdrafts, and unsecured banking credit lines.

As soon as the credit arrives, they max out and move on. U.S. banks are losing tens of billions of dollars every year. Up to 20% of unsecured bad debt at U.S. and European banks is actually fraud.

Unfortunately, this type of criminal activity is beyond the capability of many financial institutions to detect.

With billions of transactions, emails, texts, and individuals involved, detecting and preventing fraud is no small undertaking.

That’s where graph databases such as Neo4j come into play. This technology is all about unearthing insight from complex relationships that may exist within a mountain of apparently unrelated data.

Unearthing Fraudulent Connections from Data

Graph databases are built around interconnected nodes, enabling deep understanding of anomalous patterns and links that would otherwise go undetected.

Take the case of our online con artists. There may be multiple groups and individuals that execute the scam.

Neo4j has the ability to unearth commonalities, such as linked phone numbers and addresses. Even if the cyber criminals leave a labyrinth of fake identities, emails, and addresses, graph databases can discover relationships at key points during an investigation—when the account was created or when credit balances change—in order to catch a crime ring before they cash out and vanish.

This opens up a whole new way to detect fraud as it is happening via real-time analysis of data relationships.

Bank fraud, money laundering, insurance fraud, and eCommerce scams can be successfully prevented by using Neo4j to detect the warning signs that represent fraudulent activity.


Overcoming RDBMS Limitations for Better Fraud Protection

Because a regular RDBMS organizes data into tables, it is unable to detect fraud or obtain the real-time insight from data.

This makes it nearly impossible to spot links among millions of transactions. Some results can be obtained if backed up with massive amounts of compute power.

But that is a very expensive and inefficient approach.

A far better and more cost-effective way to uncover fraud rings is to harness graph databases to analyze large, complex, and highly-interconnected data sets.

Only technologies such as Neo4j can move fast enough to raise a red flag the moment a suspicious account is created and then tie it to a fraudulent transaction in real time.

Neo4j: The Standard in Fraud Detection

The bad guys move fast.

Their crime rings are continuously evolving. They grow in shape and size and then disappear without trace in minutes.

Therefore, a fraud detection application has to be able accommodate highly dynamic environments and reach the right conclusions almost instantaneously.

Neo4j uncovers these difficult-to-detect patterns. Rather than relying on traditional representations such as tables, it stores interconnected data that is neither linear nor purely hierarchical, making it easier to detect malicious behavior regardless of the depth or the shape of the data.

When supported by a native graph processing engine that supports high-performance graph queries on large datasets, Neo4j quickly becomes the standard in real-time fraud detection.

Find out more about new ways to detect fraud using Neo4j.

Written by IBM BP Network