I've been meaning to play with the twitter streaming api. There are so many ways to explore applications of it.
With node.js, there are a ton of available modules that make it very simple to put together an app that uses the streaming api.
So, last night and this afternoon I finally started putting together some pieces, tweaking on features of a budding app. For what I do not know - I am just seeing what might naturally arise.
The current specific pieces are:
- twit - for handling gruntwork for dealing with both the REST and streaming twitter interfaces
- retext-sentiment - for (trying to) detect the sentiment of text
- say.js - for having my mac speak tweets that are flying by, randomly choosing one of the built-in voices for each tweet (tweets are spoken only if one is not currently being spoken)
- smslingo - for converting "sms speak" to their multi-word equivalents
A while back, I had looked into sentiment analysis and was a bit disappointed. I believe that this may still be true - at least for real-time. Of course, it might not matter in the large if you've got a firehose of tweets coming at you.
Last night, I was listening to the rush of tweets about Michigan after that once-in-a-lifetime game-ending play. This afternoon, it's been the NFL. Just silly chatter, often very funny. There is something oddly soothing about the ongoing activity of life.