My Bachelor's thesis on Corruption Risk Profiling with Open Data. Explores application of graph theory and anomaly detection ML to detect corruption risks, based on Ukrainian state register of e-asset declarations of public officials (NACP/НАЗК data).
See the full thesis here: https://github.com/ipopovych/open_data_networks/blob/main/Thesis%20Iryna%20Popovych.pdf
One of the primary obstacles to the effective prevention of corruption is the lack of knowledge and understanding of the circumstances under which corruption arises and evolves. In this work, we develop an approach for open data processing aiming to build network models of public officials in Ukraine and identify relations between them as an indicator of corruption risk. The primary data source for this research is the Unified State Register of Declarations of Persons Authorized to Perform Functions of the State or Local Self-Government. We propose to single out the facts of social and economic relations between the subjects of the declaration (public officials), and on this basis, to model the networks using principles of social network analysis and graph theory. Our results show that the social networks of Ukrainian public officials can be modeled by defining relationships between them via company co-ownership or common participation in an organization. We explore the resulting networks with modularity coefficients, review communities within networks, and propose a framework for corruption risk profiling with the usage of network characteristics.
“We cannot understand our humanity just by studying individuals.” -- Nicholas A. Christakis