I'm trying to write a blog post about social-object graphs but I keep getting distracted finding new datasets to generate graphs from.
The SXSW2012 panel picker just opened and every proposal has a set of tags on it. To find out what clusters of things 2012 will care about, I generated the tag co-occurrence graph and did some analysis and layout in Gephi.
In-depth zoomable version: http://biddul.ph/pSidY0
EDIT: This visualisation was made with Gephi with the Seadragon Ajax export. I obtained the data with a bit of ruby screenscraping code that outputs a GDF file (graph format supported by Gephi). This is the resulting data file, in case anyone wants to load it up in Gephi themselves: http://www.hackdiary.com/misc/sxsw2012.gdf
More thoughts here: https://plus.google.com/114200438004839793612/posts/j7cRQHM4f9T