TouchGraph is a force-based graph layout engine that’s been around for quite a while. The free version still available on SourceForge is an old standby. There was a lot of inspiring conversation about visualizing collections and aggregate information at Museums and the Web and I wanted to see what kind of results I could get with with the data available through the Pipes.
Navigating the graph is accomplished by clicking on the nodes, and using the drop-down at the upper-left and the slider across the top of the window. Mousing over the Object nodes opens up a tooltip with the Met collection database URL. Double-clicking on any of the Identity nodes loads the relevant New York Times Identity search in a new window. Still trying to figure out which one I prefer.
I hope it’s just my own Java issues causing the applet pause in Firefox and Explorer. It works fine in Safari… let me know how it works for you…
Ah, right. Where’d the data come from? I wrote a Processing script to call the New York Times Identity Pipe, iterating over the first page of Modern Art entries in the Met’s collection database. The resulting network is made up of all of the Objects provided and all of the Identities returned by the pipe. A directed edge is added from the Object to each Identity returned by the pipe. Since some of the Identities map to more than one Object, the result is a network of the people the Times sees as related to Modern Art in the Met.
I’m not a huge fan of ball-and-stick network graphs – so much of the nuance gets obscured. I hope to try other visualizations soon.
//Todo: Remove as many manual steps from the process as possible. The Processing script runs locally on a CSV file with Object IDs. (Using JSON in Processing.) Without a more robust connection to the underlying data, automating the process would be a bit of a hack. Maybe something to try with the Brooklyn API? It’s not any trouble to pull a lot more data using Processing, but I wonder how well TouchGraph performance scales?