Computing the statistical significance of optimized communities in networks

07/09/2018

∙

by   John Palowitch, et al.

∙

0

∙

share

It is often of interest to find communities in network data as a form of unsupervised learning, either for feature discovery or scientific study. The vast majority of community detection methods proceed via optimization of a quality function, which is possible even on random networks without communities. Therefore there is usually not an easy way to tell if an optimized community is significant, in this context meaning more internally connected than would be expected under a random graph model without true communities. This paper introduces FOCS (Fast Optimized Community Significance), a new approach for computing a significance score for individual communities.