Predictive Multigraph Analysis
01.02.2020 – 31.01.2023
In the field of fraud prevention, the procedures of predictive analytics represent an essential and established approach. A subset of these procedures deals with methods of Social Network Analysis (SNA), which is based on graph theory. Such methods are used in a wide variety of fields and industries, such as telecommunications, banking or in the fight against disinformation.
A graph data structure is built from relationships between actors observed in the real world. In the field of combating financial fraud, these can be employee migrations (pay slip data), contract awards (client liability in the construction industry), joint suppliers in third countries (customs declarations), participation structures (company register & WiReg) or acting persons (company register).
The current state of science knows a number of methods for fraud prediction based on individual graphs. However, as in the case of the financial administration application described above, there are usually different sources of information available that cannot be modelled as a graph without losses. Such networks are therefore modelled as so-called multigraphs, which depict different types of relationships in a network.
For these multigraphs, however, there is a lack of established methods for fraud prediction. The aim of this project is to develop methods for fraud prediction that can be efficiently applied to large-scale multigraphs.
The Research Team
