This is more on the experimental side of "research" but I just
finished a prototype realtime visualization of tweets that reference
Wikipedia:
http://wikitweets.herokuapp.com/
wikitweets is a NodeJS [1] application that listens to the Twitter
Streaming API [2] for tweets that contain Wikipedia URLs, and then
looks up the relevant Wikipedia article using the API to ultimately
stream the information to the browser using SocketIO [3]. The most
amazing thing for me is seeing the application run comfortably (so
far) as a single process on Heroku with no attached database needed.
If you are curious the code is on GitHub [4].
The key to wikistream working at all is that Twitter allows you to
search and filter the stream using the original (unshorted) URL. So
for example a Tweet with the text:
Question of the Day: What’s the greatest seafaring movie ever?
Some suggestions:
http://bit.ly/IqsE1e (But anything on water'll work)
#QOD [5]
Is discoverable with a search query like:
Question of the Day
wikipedia.org [6]
Note "wikipedia.org" doesn't exist in the text of the original tweet
at all, since it has been shortened by bit.ly -- but it is still
searchable because Twitter appear to be unshortening and indexing
URLs. Anyhow, I thought I'd share here since this also relied heavily
on the various language Wikipedia APIs.
//Ed
[1]
http://nodejs.org
[2]
https://dev.twitter.com/docs/streaming-api/methods
[3]
http://socket.io
[4]
https://github.com/edsu/wikitweets
[5]
https://twitter.com/#!/EWeitzman/status/195520487357558784
[6]
https://twitter.com/#!/search/realtime/Question%20of%20the%20Day%20wikipedi…