Graph databases

back to the product groups

Graph IQ

The structure of the Graph IQ tool allows for adaptation the data structures to an ever-changing requirements (e.g. of the anti-fraud strategies).

Fraud detection

Financial frauds are getting more and more difficult to trace. The fraudsters are creating:

artificial identities; based on real data from different sources new identities are being forged
connection networks, both small and big. The greater number of connected persons and entities, the bigger scale of alleged frauds

Standard mechanisms of fraud detection (also those Machine Learning based), do not – to a great extent – discover the situations mentioned above; mainly due to the fact that analysed are the data, not the connections. If we connect persons using the same phone number or ID number, using the same phone number and email address or persons who share the same home address, it will turn out that they will receive the status of „related” – although in traditional databases they will remain stranger to each other. Algorithms within the graph database allow for discovering connections and information groups, which – in turn – helps in discovering fraud chains.

Support in the management of the concentration risk

Defining adequate conditions, like:

the same kind of activity
dealing with the same kind of goods
the same issuer of a security
connections within the capital groups
identification of the control relations (e.g. preparing the consolidated reports, right to vote, decision making power)
identification of the economic dependency (e.g. mutual warranty, siginificant connections between the client, supplier and receiver, common owners etc.)

we can create a clear graph, containing nodes and relations within previously indicated transactions and conditions, and subject them to further analyses. It is also possible to use the data from the external sources.

We can introduce to the graphs persons and relations not resulting directly from the court registers, e.g. kinship, neighbourhood, close co-operation of the companies not related formally, transactions within the supply chains etc. With further analysis we can use these relations or we can leave them out.

Making additional relations can be done by defining the rules for data in the base and their implementation. These data may come from the internal base of a bank (e.g. relations referring to a bank account of a specific person) or from external sources.

There are also algorithms predicting certain ways of behaviour of some persons, according to the behaviour of their „friends” (i.e. persons directly related). Defining interpersonal connections may be of a great significance – now, and in the future.