Visualizing Twitter clusters

I’ve seen lots of maps of online social networks. They’re nifty, but often either ugly or messy or useless. But I came back from Strata really impressed with LinkedIn’s InMaps. The layout of the maps is pleasing, but more importantly, they use cluster analysis to demonstrate something that’d be harder to do if not visualized. If you have a LinkedIn account and haven’t tried InMaps, do.

In InMaps, clusters tend to represent companies or communities-of-interest. I wondered what they might represent if I fed it the network of people that I follow on Twitter.

More on why I think this might be interesting: I’m going to be looking for a job soon, in a field that’s different than any that I’ve worked in before. While my work represents what I’ve done and been interested in in the past, the set of people I follow on Twitter are more representative of what I’m interested in now. Indeed, I spend a sometimes-embarrassing amount of time keeping up with what’s happening in these communities. I do it because I get value out of knowing what’s new & important. In some cases, this currency is something that a potential employer would value too. I don’t yet know how to articulate this value to potential employers. I figure that by exploring this a bit more, I might come to something. (There lots more here; I’ll keep thinking about it and explore it some more in a future blog post.)

So, I hacked together some php scripts to pull data out of the Twitter API (though, as it turns out, the Google Social Graph API does what I need for now just fine, and is faster and simpler), and plugged it into gephi. In my first few tests, here’s what came out:

They’re cool, but they’re not exactly what I was looking for. I’ve got a more recent batch which I’ll post shortly, and I have a few new ideas which I’ll try out soon.

It didn’t take long for me to hit the limits of what my otherwise-capable MacBook could handle. I’ve since found an 8-core Mac Pro with 10GB of RAM (thanks, OCADU!), and once I got gephi working in 64-bit mode, things got more comfortable.

FWIW: while trying to troubleshoot some problems with gephi, I found Tony Hirst’s blog. He’s been doing similar work for a while now. <respect>

This entry was posted in infoviz. Bookmark the permalink.

Leave a Reply

Your email address will not be published. Required fields are marked *