Greg Morrison

Greg Morrison


Harvard University

School of Engineering & Applied Science

Wednesday, February 23, 2011

4:30 p.m.

Cardwell 102

Generalized Erdos Numbers, Networks, and Communities


Concepts from graph theory have been applied to a wide range of physical, biological, and sociological systems.  In this talk, I will present a novel topological measure of closeness between nodes, the Generalized Erdos Numbers (GENs).  While the approach is derived with collaboration networks in mind, this measure can be broadly applied to a wide variety of problems.  The utility of the GENs will be explored by considering a variety of real world and artificial examples, such as movie ratings prediction using a small subset of the NetFlix Prize data, as well as in predicting the asymmetries in first-passage times of random walks.  This topological measure also suggests a new and accurate method of detecting hierarchical communities in structured networks which does not rely on null models or arbitrary parameters.