10 Social network analysis against fraud

 

This chapter covers

  • Using social network analysis (SNA) to classify fraudsters and fraud risks
  • Describing different graph algorithms for SNA-based fraud analytics
  • Using a real graph database to perform a proper SNA

In this chapter, you will learn about techniques for combating fraud that approach the task from a different perspective. The techniques for fighting fraud presented in chapters 8 and 9 use different graph construction methods to create networks based on the information available in the transactions themselves and/or the users’ accounts. In chapter 8, we created a graph connecting users with merchants by using transaction information, and we explored connecting nodes based on overlapping information (two accounts that use the same email address, for example). In chapter 9, you learned how to construct a graph (the k-NN graph) by computing distances between pairs of observations (each of which has been converted to a node) and storing the top k relationships.

10.1 Social network analysis concepts

10.2 Score-based methods

10.2.1 Neighborhood metrics

10.2.2 Centrality metrics

10.2.3 Collective inference algorithms

10.3 Cluster-based methods

10.4 Advantages of graphs

Summary

References