AI for Higher-Order Graph and Network Data Analysis

Traditional computational models represent networks as a bunch of two-way connections, like lines drawn between all the cities on a map. These pairwise relationships are encoded by edges in a graph. But real-world networks often involve relationships that take place among more than two entities at once: People communicate in groups, students learn from each other in classrooms, and biological interactions occur among sets of molecules rather than between only two. Attempting to understand networks as a collection of pairwise relationships limits traditional network models. Going beyond graph methods is necessary to fully realize the richness of the higher-order interactions in complex data sets.

With this CAREER award, Austin Benson, Computer Science, is developing new artificial intelligence (AI) capabilities for analyzing the massive data sets that arise from higher-order connections in real-world networks. The research will focus on three models of higher-order network data: hypergraphs, tensors, and simplicial complexes. For each model, Benson will develop new data-mining and machine-learning methods and apply them to a variety of data to demonstrate applications in social networks, e-commerce, biomedicine, and information management.

This project aims to introduce the next generation of network data analysis. The frameworks developed by this research have the potential to improve computational models for forecasting disease spread, proposing new types of drug combinations, and enhancing our ability to detect adversarial and malicious groups on the internet.

Cornell Researchers

Funding Received

$417 Thousand spanning 5 years