Analysis of Community Evolution in Networks

• Activity Number: 498
• Type: Contributed
• Date/Time: Wednesday, August 3, 2016 : 8:30 AM to 10:20 AM
• Sponsor: Section on Statistical Learning and Data Science
• Author(s): Giuliana Pallotta* and Goran Konjevod
• Companies: Lawrence Livermore National Laboratory and Lawrence Livermore National Laboratory
• Keywords: network analysis ; stochastic block models ; community discovery ; dynamic networks ; transportation networks ; change detection

Abstract

We are interested in analyzing change in dynamic networks. More precisely, we consider the changes in the activity distribution within the network, in terms link existence and intensity link weight. Detecting change in univariate metrics reduces to statistical process control. Detecting change in larger-scale structures is more challenging and less well understood. We study the problem of classifying nodes by their role in the process underlying the network, as well as detecting changes in networkstructure. Using a dataset covering all taxi trips in New York City since 2009, we investigate the evolution of an ensemble of networks under different spatio-temporal resolutions. We identify the community structure by fitting a weighted stochastic block model. We study node ranking and clustering methods, their ability to capture the rhythm of life in the Big Apple, and their potential usefulness in highlighting changes in the underlying network structure.

This work performed under the auspices of the U.S. Department of Energy by Lawrence Livermore National Laboratory under contract DE-AC52-07NA27344.