Graph analytics uniquely suited for fighting financial crime

Graph analytics can be an essential tool in the fight against financial crimes.

That was the message of Heather Adams, managing director of resilience and risk trust at Ireland-based consulting firm Accenture, who spoke on April 21 during Graph + AI Summit, an open virtual conference hosted by graph analytics vendor TigerGraph.

Fraud, money laundering and corruption, among other financial crimes, plague organizations of all kinds, and graph analytics is uniquely suited to detect such criminal activities. Other financial crimes such as terrorist funding, meanwhile, have broad societal implications.

But using graph databases, which are at the core of graph analytics, organizations can be better equipped to detect financial crimes than if they used traditional relational databases.

Graph databases enable data points to connect with one another in different ways than relational databases, making them better at discovering relationships between data points that might not be discoverable -- or would take significantly more time and effort to discover -- in a relational database.

In graph databases, data points are able to connect to multiple data points simultaneously. In relational databases, meanwhile, data points are only able to connect with each other one data point at a time. Graph databases, therefore, are better at revealing entire networks of connections.

Social media networks such as Facebook and LinkedIn, for example, use graph databases to find connections between people. And another use case, of course, is detecting financial crimes.

"It's about playing a meaningful role in society in preventing terrorist financing and stopping criminals' gain from activities like drugs and human trafficking," Adams said.

Beyond the societal role they can play in fighting financial crime, organizations have a legal responsibility to combat financial crime to prove compliance and risk management.

Financial services institutions, for example, are required to file reports with government agencies such as the FBI in the U.S. and the National Crime Agency in the U.K. any time they detect potential money laundering, terrorist funding or other financial crimes.

Data, meanwhile, is important to that fight.

And just as data needs to be cleaned and prepared before it can be used to inform strategic decision, it needs to be accessible in order to detect criminal activity. With augmented intelligence and machine learning capabilities, organizations can use data to automatically check for signs of fraud, money laundering, terrorist financing, sanctions breaches and other financial crimes, according to Adams.

Graph analytics, meanwhile, provides the best means of understanding everything possible about any customers who may be engaging in criminal activity, she continued.

It's about playing a meaningful role in society in preventing terrorist financing and stopping criminals' gain from activities like drugs and human trafficking.

"Graph technologies can connect that data really effectively across the internal and external data sources and identify relationships between parties," Adams said. "This can enable you then to look at the risk associated across those relationships rather than just looking at data regarding the current party alone."

Natural language processing is one of the AI capabilities organizations can use to help detect financial crimes. With unstructured data, NLP can pick up information from media reports and documents, finding links based on key words and phrases.

Machine learning, meanwhile, enables organizations to sort through unstructured data in an efficient way that produces meaningful results that can then be scored for their potential risk and fuel decisions about whether they merit investigation.

"We can use analytics to look at risk factors within a network of relationships and transactions between people and companies," Adams said. "This is where we look at the network view, and we can use graph technologies to better understand connections."

Beyond scouring data for suspicious activity, organizations can use AI and ML to establish patterns of normal behavior against which potential criminal activity can be compared, and even for a specific person or organization compare current behavior against past behavior.

That, according to Adams, can help remove false positives and enable organizations to home in on truly suspicious activity. Financial institutions, in particular, struggle with false positives, making truly suspicious transactions difficult to identify.

"Tools like graph analytics can be really helpful to actually visualize the connections between different parties, complex hierarchies and different ownership structures, presence and movement of business activities across geographies," Adams said. "Building that out in a network view that looks at connections can really help a person work a case."

And potentially stop criminal activity before real harm is done.

"Any step forward in better identifying where fraudsters are taking advantage of customers, or where your organization is being abused by criminals, is a great step forward and one that society, as a whole, can benefit from," Adams said. "Data and analytics can absolutely be part of that journey, and can be really important tools."