Department of Computational Social Science Seminar-Gutfraind

Thursday, February 28 - 1:00 p.m.
Center for Social Complexity Suite
Research Hall, 3rd Floor

A Proposal for Simulating Social Networks

Alexander Gutfraind, Ph.D.
Center for Computational Biology and Bioinformatics
The University of Texas at Austin

ABSTRACT: Social networks fuel and steer most social processes, and for this reason they are an important component of many simulations. When simulating social networks, two of the most common approaches are to construct them mechanistically and to use a stylized model from graph theory. Unfortunately, these approaches do not always adequately model the complex topologies and correlations of real social networks. In this talk, I will propose a new approach for simulating networks, termed MUSKETEER. MUSKETEER is initialized with a prototype network, and uses a series of mappings that coarsen and later refine the network structure, while introducing an arbitrary amount of variability at multiple scales. MUSKETEER is designed to accurately simulate many types of network properties, to preserve even unknown or unspecified features, and to give realistic annotation of nodes and edges. Using examples from several domains, I will show that MUSKETEER indeed produces the intended variability while achieving greater realism than existing algorithms.

Joint work with I. Safro and L.A. Meyers