MIT Technology Review covered an effort by developer Gilad Lotan that visually charted the spread of Twitter buzz caused by the recent unrest in Iran. According to the article:
During the Iranian elections in June, microblogging site Twitter became a
way for protesters to communicate with each other and with the rest of the
world. Stories of oppression, police brutality, and violence spread via
140-character tweets despite the government’s efforts to filter
Web content and control
Web traffic inside the country.
A new visualization tool developed by Gilad Lotan, a programmer and designer at Microsoft Startup Labs, shows just how information related to the elections spread through
Twitter, via the most popular Twitter conversations and retweets.
I found this to be fascinating on a number of levels. It illustrates the ability of social media to play a starring role in events (and influence them) as they unfold. Also, the tool he developed shows what is possible when it comes tracking the spread of buzz.
I have been blogging about improved tools for tracking online info, and the story definitely caught my attention. So I sent Gilad an email and had the following exchange with him. Please read and enjoy. And feel free to comment if you are aware of any metrics, analysis and reporting tools for Twitter.
Q (Bob Geller): I know your effrot is a work in progress and possibly won’t be packaged or commercialized. Are you aware of any existing commercial products or Web utilities that report on and help analyze Tweeting and Retweeting activity?
A (Gilad Lotan): My work is definitely in progress and will not be commercialized. I’m looking at the possibility of releasing the code to other developers or creating an online service where people can subscribe to personalized queries, which would be free. However, I would need to change the UI quite a bit, as this project is designed to fit the Iranian Elections content. An actual tool would need to be much more functional, and probably not written in JAVA.
There exist a variety of web “trending tools” that follow twitter and highlight retweeted messages (i.e. http://www.retweet.com/), however none that I found do the type of analysis that my visualization enables. None make matches between similar texts and highlight differences in words authors used to portray the same message.
I imagine that with the upcoming rollout of Twitter retweet API support, it will be substantially easier to follow retweet trails to the originating tweet, and have similar analysis, without the need for the more complex set of Natural Language Processing queries that I am performing.
Q: I found a site called Retweetist.com that could help too. In addition to the advanced functionality you cite – visualization, NLP content matching – it would be great to find a dashboard that enables basic Twitter metrics, and provides reports on who replies, Retweets, how many followers they have, etc. to see the reach of Tweets, replies and RTs without jumping through hoops. Are you aware of any such tool?
A: I am not aware of a tool similar to what you’re describing. I feel like I have stumbled upon subscription based tools that do some metrics/analytics, but none attempt to tackle any of the NLP problems.
