I pushed out a little change to my Big in Twitter toy. (Recap: it sucks in lists of band names and then searches Twitter for them. When it finds a match it’ll then grab the Twitterer’s location. So at a glance, see what (and where) bands are getting the most mentions.)
This version adds some basic classification to the search results. If your band name is a common English word (today has a good example, the band “Sunset”) Twitter will have a lot of search results. I use the classifier gem to decide if the result is related to music or not. For the most part it does the job but the results are much better the more unique the band name. This is an area I’d like to research further and a subject Ruby doesn’t necessarily have the libraries and history for….
The only downside I’ve run into so far is keeping the search code running nonstop. Much much better uptime on my Mac than on my Ubuntu machine. The culprits tend to be hpricot (having to use a slightly out of date version (v0.6) due to my older Ubuntu) or ruby itself. The ruby crashes are pretty spectacular actually.
So Cold Pumas next time you’re in Brighton look up danielnixon, he’s a fan!