Starting with the Michael Bommarito’s twitter handle mjbommar, we built this visualization by collecting Mike’s direct friends, friends-friends, friends-friends-friends, etc. until we decided to stop …. just after passing 130,000 total twitter handles. Using the Fruchterman-Rheingold algorithm, we visualized a network where |V| = 130365, |E| = 197399.
Those interested in reviewing some other twitter visualizations, please consult Nathan Yau at Flowing Data who has collected some of his favorites. To our knowledge, the visualization we offer above is one the larger visualizations of twitter that have been produced to date. When you zoom in, you will notice we have flagged some of the celebrity twitter users we detected in the mjbommar friends-friends-friends, etc. network. For example, as shown above Ashton Kutcher (aplusk), Chad Ochocinco (OGOchOCinco) and RainnWilson (rainnwilson) are contained therein.
Given the budget limitations of this blog, we cannot host this visualization in house. However, if you click the picture above, you can access the visual from Seadragon … a zoomable visualization platform from Microsoft Labs.
3 thoughts on “Large Scale ( 130,000 + ) Zoomable Visualization of a Twitter Network”
Generating a FR layout for 130365 nodes must have taken quite some time. What software did you use (igraph?), and on what hardware did it run?
The calculation was done with a modified version of igraph. A couple hundred iterations run in just over a day on a new i7.
And to think I was going to talk to soemone in person about this.