Advancing Model Performance and Forward Selection Algorithms in Network Science: Applications for Criminal and Corporate Law Enforcement

This project develops novel network science methods to better analyze and predict complex systems. The resulting models are applied to burglary prediction, cartel detection, and legal citation networks, creating practical data-driven tools for law enforcement, competition authorities, and legal research.

Key data

Description

This project develops novel methods in network science to improve the analysis and prediction of complex interconnected systems. It introduces new approaches for evaluating network models and selecting informative features, enabling more reliable and interpretable graph-based machine learning.

These methods are applied to three real-world domains: burglary prediction, cartel detection, and legal citation networks in European competition law. By combining mathematical innovation with large-scale data analytics, the project aims to support evidence-based decision-making for law enforcement agencies, competition authorities, and legal practitioners.

The resulting open-source methods and software will advance applied research while providing practical tools for crime prevention, competition policy, and the analysis of evolving legal systems.