Technologist Emil Eifrem says a combination of graph databases and the Person, Object, Location, Event data model trialled in Greater Manchester exposed a web of networks...

Graph database technology is a powerful way of both recognising and leveraging connections in large quantities of otherwise random data. The International Consortium of Investigative Journalists used it to detect fraud and corruption in its recent famous Panama and Paradise Papers global probes, for instance, while Google uses a graph-based way of representing knowledge to enhance its search engine and map the Web.

Whether it’s more effectively guarding borders, spotting possible terrorist activity, detecting welfare fraud or helping uncover scams and hacker attacks, graph databases are a powerful enabler for getting ‘big picture’ connected analysis. But unlocking useful insights based on connections, graph technology may also offer a way to support the Police, social services and other Government agencies.

Ten years ago, a G8 country’s main immigration authority started to work with graph database technology in order to help it visualise relationships and connections. This was essentially a case management use case – a way to helping them work more efficiently with individual cases of potential interest to border control officers.

What this state actor found is that (deliberately) hidden connections become more obvious when looked at with a system designed to manage connected data, delivering the team a way to run real-time queries for detecting a variety of criminal networks or fraud rings.

Graph databases are also being looked at to help enable a new highly responsive informal learning system to support decision-making and incorporating social media.

The idea hinges on who knows who – the very definition, when you think about it, of a relationship: if person of interest X has come to the attention of the authorities for whatever reason, then who else in X’s network might be worth keeping notice of? Perhaps they are in relationships with people who have connections to other people with criminal records, for example – like an ex-offender who could slip back into unhealthy relationship patterns, say, or family members that could be potentially drawn into trouble or placed at risk.

Police practitioners are looking into this by working with the emerging POLE (Person, Object, Location, Event) data model for working with crime data. POLE is a great fit for graph database technology and graph algorithms, an affinity that can be made even more useful by linking it to data visualisation front-ends, including popular tools like Tableau.

Our team recently tested this hypothesis – and found some really promising result. Using a sample public dataset of one month’s worth of Greater Manchester street-level crime the researchers cross- connected a number of other data sources, from geotagging data to addresses to randomly generated person information to see how deep a picture of these connections could be generated. An interesting web of networks soon became exposed, mapping 29,000 crimes in 15,000 locations, generating 106,000 relationships between the nodes.

You can explore the results of this proof of concept on our YouTube video (see the Custodial Review website), but what this and other examples of graph as a public sector aid shows us is that graph database software and crime data could be a really potent combination, enabling data-driven investigations and decision making, and allowing police forces and law enforcement and Government agencies to intelligently maximise their resources in the face of budget constraint and an ever- evolving landscape of crime and security threats.

• Emil Eifrem is co-founder and CEO of Neo4j, the world’s leading graph database.