State registration number: 0120U102151

Period of execution: 2020 - 2022

Source of funding: funds from the State Budget of Ukraine (Order of the Ministry of Education and Science No. 29, dated 11.01.2020)

The aim of the project is to model risks and security in the digital economy based on a new concept based on Data Science models and methods, and to develope an intelligent system for automated construction of scoring models to assess the risks of improper receipt of social assistance and tax evasion.

Main tasks:

1. Develop a new concept for risk and security modeling in the modern conditions of economic digitalization, in particular:
2. Analyzing the digitalization processes occurring in Ukraine’s economy and classifying risks and threats in the digital economy.
3. Examining changes in employment structure and population income under digitalization and their impact on tax evasion risks and unlawful receipt of social benefits.
4. Analyzing issues related to the protection and security of commercial information and personal data in the digital economy.
5. Conducting an analytical assessment of existing mathematical tools for modeling risks and security in the digital economy and identifying the most suitable tools for adequately modeling tax evasion risks and fraud in social benefit claims.
6. Enhance the toolkit for risk and security modeling in the digital economy, which includes:
7. Investigating the behavioral models of economic agents regarding tax payment in a digital economy.
8. Defining prior segmentation of taxpayers and selecting appropriate mathematical tools for modeling tax evasion risks for each segment.
9. Analyzing the behavioral models of economic agents in obtaining social benefits in the context of economic digitalization and the spread of remote employment.
10. Defining the segmentation of social benefit recipients based on prior information and selecting suitable modeling tools for assessing the risks of unlawful social benefit claims for each segment.
11. Develop an intelligent system for automated construction of scoring models to assess the risks of unlawful receipt of social benefits and tax evasion using Data Science methods and models (Data Mining, machine learning, including artificial intelligence methods), which involves:
12. Developing software for implementing scoring models to assess tax evasion risks.
13. Developing software for implementing scoring models to assess the risks of unlawful receipt of social benefits.
14. Training neural networks on real-world data.
15. Perform testing and debugging of the developed software.

Outcomes of the project:

• A new concept for risk and security modeling was developed, taking into account global economic digitalization processes occurring in Ukraine and worldwide, corresponding changes in employment structure and population income, and their impact on the risks of tax evasion and unlawful receipt of social benefits, as well as cybersecurity issues in the digital economy.
• A series of economic and mathematical models were built, and corresponding intelligent systems for risk analysis and automated construction of scoring models for assessing the risks of unlawful receipt of social benefits and tax evasion were developed. These systems enable the uploading of historical data and the adjustment of models to identify characteristics typical of compliant taxpayers and social benefit recipients, as well as law violators.
• The system operates automatically, performing the following tasks:
• Dividing the overall sample into training and test sets with a balanced representation of different taxpayer or social benefit recipient behaviors.
• Categorizing taxpayer or social benefit recipient characteristics while ensuring systematic consideration of the influence of categorized quantitative indicators on the target variable.
• Selecting characteristics that provide the highest efficiency and stability of assessment results on both training and test samples.