Exploiting data analytics for social services: on searching for profiles of unlawful use of social benefits

In this paper we present a data-driven profiling approach that we have adopted and implemented for a municipality. Our aim was to make profiles transparent and meaningful for citizens, policymakers and authorities so that they can validate, scrutinize and challenge the profiles. Our approach relies on a Genetic Algorithm (GA) that searches for useful and human understandable group profiles. Furthermore, we discuss some of the challenges encountered, show a selection of the profiles that were found by the GA, and discuss the necessity and a number of ways of validating these profiles in accordance with, e.g., privacy and non-discrimination laws and guidelines before using them in practice.