Large Social Network Dynamics Talk + Sunbelt Presentation

If you are curious about new approaches to studying large social network dynamics, check out the confirmed upcoming talk at the International Network for Social Network Analysis Conference (Sunbelt) in Hamburg, Germany. I will be giving a talk, on behalf of my co-authors (see below), on the “Methodological Specifications for Application of the Mean-field Model for Large Scale Social Networks” in the section- Large Scale Networks Analysis on Thursday afternoon 23 May 2013. In this talk I will discuss a model we have developed for overcoming a number of limitations in presently used models to investigate large social network dynamics.

For further details, as they become available check out – the conference webpage:

Submission Summary:

Title: Methodological specifications for application of the mean-field model for large scale social networks
Author(s): Birkholz, Julie M1, Lungeanu, Alina,2, Bakhshi, Rena3, Groenewegen, Peter1, van Steen, Maarten3, Contractor, Noshir2
Institute(s): 1Vrije Universiteit Amsterdam, Network Institute, Organization Sciences, Amsterdam, Netherlands, 2Northwestern University, Evanston, IL, United States, 3Vrije Universiteit Amsterdam, Network Institute, Computer Science, Amsterdam, Netherlands

Text: The statistical modeling of the emergence of social networks is most commonly undertaken using two models: Stochastic actor-orient models (using SIENA) and p*/ERGM. However, both models have scaling limitations due to computational challenges. We propose the use of a mean-field approach to study large scale social network dynamics (1000s nodes). A mean-field model, originating from physics, enables consideration of a large number of nodes through the aggregation of classes of nodes into “nodal buckets.” The analysis then computes the interactions/communication between buckets. The mean-field model has been successfully applied to estimate attribute and network parameters on large social networks (Birkholz et al 2012). Here the nodes were aggregated into buckets based on shared attributes.

However, the inferences from such models hinges crucially on the selection of the shared attribute used for classification of nodes into buckets. To overcome this limitation, we propose a methodological specification (based on equivalence classes) for the aggregation of nodes into buckets. We apply this technique to study the the co-authorship network of 1,354 researchers in the Oncofertility scientific field over a four year period. We estimate the extent to which collaboration networks are influenced by multiple factors such as cosmopolitanism, visibility, scientific age, and institutional affiliation.


Studying Large Social Networks

In my research I investigate dynamics in large social networks, networks of >1000 nodes. The most commonly used models have a number of limitations for studying the combined effects of both network and social parameters on network evolution; thus a year ago I started working with Rena Bakhshi, a very talented researcher within the Network Institute. Rena had a model- the mean field model, which had used to investigate dynamics of large communication networks and wanted to experiment with social network data. This resulted in our first application of the mean-field model for large social networks, entitled – Scalable Analysis for Large Social Networks: The Data-Aware Mean-Field Approach, recently published in Social Informatics. See the abstract here.


Studies on social networks have proved that endogenous and exogenous factors influence dynamics. Two streams of modeling exist on explaining the dynamics of social networks: 1) models predicting links through network properties, and 2) models considering the effects of social attributes. In this interdisciplinary study we work to overcome a number of computational limitations within these current models. We employ a mean-field model which allows for the construction of a population-specific model informed from empirical research for predicting links from both network and social properties in large social networks.. The model is tested on a population of conference coauthorship behavior, considering a number of parameters from available Web data. We address how large social networks can be modeled preserving both network and social parameters. We prove that the mean-field model, using a data-aware approach, allows us to overcome computational burdens and thus scalability issues in modeling large social networks in terms of both network and social parameters. Additionally, we confirm that large social networks evolve through both network and social-selection decisions; asserting that the dynamics of networks cannot singly be studied from a single perspective but must consider effects of social parameters.