EAGER: A Graph Analytics Approach to Understanding Leakage Patterns in Information Espionage

Scientific research often relies on open collaboration across borders, institutions, and disciplines. But in today’s competitive global environment, there is growing concern that valuable discoveries funded by U.S. taxpayers may be subject to malign foreign interference or misappropriation. This project develops new tools and techniques to help universities and research agencies better understand when and how such transfers of knowledge might occur. By analyzing how people, publications, and ideas are connected, the system will flag unusual patterns—such as researchers moving abroad and using U.S. research without attribution or foreign patents that closely resemble federally funded work. The long-term goal is to help institutions protect research integrity while continuing to support open and collaborative science. Technically, the project constructs a knowledge graph that connects data from grants, publications, patents, and institutional affiliations over time. Within this graph, it defines and detects “leakage paths”—chains of connections that suggest research may have moved outside authorized channels. The system uses a combination of human-defined templates and machine learning techniques, such as temporal graph neural networks, to score and explain these patterns. Simulated scenarios of potential misuse are also used to improve the models and discover new risk indicators. The final output, a proof-of-concept system, will be shared as an open-source toolkit, including a searchable API, interactive interface, and reproducible simulation environment. If successful, the project will provide a new kind of research security infrastructure—one that is proactive, explainable, and aligned with evolving national priorities. This award reflects NSF's statutory mission and has been deemed worthy of support through evaluation using the Foundation's intellectual merit and broader impacts review criteria.