figshare
Browse
lbauguitte_conpast_april2014.pdf (492.04 kB)

Models of historical networks: A methodological proposal

Download (0 kB)
presentation
posted on 2014-06-04, 12:47 authored by Laurent Beauguitte, The connected past The connected pastThe connected past The connected past

Network analysis is more and more currently used in historical and archeological studies and several recent initiatives in France (RES-HIST), United Kingdom (Connected Past) or Germany (Historical Network Research) illustrate this trend. However, most studies remain strictly empirical: the aim being to characterize a specific network or some index regarding given actors. If this first step is necessary, it does not allow comparison nor generalization. Our objective is to propose a way to gain generalization regarding historical network analysis and to provide basis for comparison.

One alternative from the data-driven process would be to create ideal models of networks derived from expected behaviors of actors. In other words, we should be able to compare our observed networks with networks created by a given process.

In my PhD regarding states and regional groups behaviors at the United Nations General Assembly (Beauguitte, 2011), I perform network analysis in a “classical way”: data collection, definition of (hopefully relevant) periods of observation, measures and interpretation. My main hypothesis regarded a possible political world regionalization: states should act more and more through regional groups at the UNGA. An alternative way would be to draw (and measure) the network expected if this regionalization process was complete.

The following figures regard speeches habits at the UNGA: most speakers are state representatives (yellow circles), but regional groups also talk (blue squares), and state representative can support group declarations. In this last case, it can be represented by a 2-mode network state-group, a link indicating that the state a supports the declaration made by the group A.

The sequence below is completely hypothetical. The first step shows a purely stato-centric situation: groups talk but states do not support (density equal 0) and the last step would be an optimum regarding regionalization: only regional groups talk (and density remains at 0).

Between these two optimal situations, b shows a group vs group configuration, c a limited regionalisation process, leaving apart some states and groups and d shows a more complex picture where one state supports two regional groups. These small figures are basic representations of stylized facts, but all can be characterized by some given measures (density, number of components, degree distribution etc.) and could be used as references when dealing with actual data.

L. Beauguitte, 2011, L’Assemblée générale de l’ONU de 1985 à nos jours : acteur et reflet du Système-Monde, Thèse de doctorat, Université Paris 7, available on Thèses en ligne.

History

Usage metrics

    Licence

    Exports

    RefWorks
    BibTeX
    Ref. manager
    Endnote
    DataCite
    NLM
    DC