Tuesday, March 5, 2013

Social network analysis with Gephi

Social Network Analysis or (SNA) is one of the most powerful tools available in the Social Sciences today. It allows to identify key players in a system, find structural holes and many other aspects of a society which are less accessible without the network paradigm.

Gephi is an opensource package written in Java which is aimed at performing SNA. I came upon this in the recommendation of the new Coursera course on SNA. Gephi was selected due to its many features, because it is free software and since it cross platform.
Another of it its advantages is in visualization.
Here is an image of a classical data-set showing the dining school partners. This is a directed graph with 26 vertecies and 52 nodes indicating each girl's top two preferences for a dinner partner. The data was collected circa 1905 and published in the 1950. The outgoing degree is therefore 2 but the incoming degree varies from 1 to 6 showing that some girls (Eva and Marion indicated in cyan) are more popular than others. The graph shows the nodes color colored by incoming degree. Gephi also provides a legend for this graph but it is not so clear how to export it.
A sociogram of the dining table preference Data set
There are a number of layout options and it is fairly easy to adjust the label size or move node around.
The missing legend showing incoming degree color code and its distribution 
I got this final chart by a screen cap