Emilie Purvine, North Pacific National Lab

Abstract

Network science has dominated analysis of complex relational data for decades. This is the practice of modeling data using a graph to represent pairwise relationships and then applying graph theoretic concepts--including degree distribution, diameter, centrality, and clustering--to understand the large scale structure of the graph/data. Network science has proven useful in a variety of domains including cyber security, bibliometrics, and computational biology. But in many of these cases the pairwise relationships that comprise the graph are inferred from more complex multi-way relationships. For example, a paper with more than two authors or a biological protein complex with more than two proteins. In these cases of multi-way relationships a hypergraph is a more accurate model of the data. In this talk I will describe the work my colleagues and I have been doing to develop theory, software, and use cases for hypernetwork science. I will introduce the relevant definitions and theoretical results to generalize network science concepts to hypergraphs and will close by showing some real examples using our HyperNetX Python package.

Presented via Zoom: https://rutgers.zoom.us/j/94346444480

Password: 6564120420