In tragedies and big events, Twitter is amazing. When the MIT shootings and the chase that followed, Twitter immediately was filled with millions of tweets that kept going and going for dozens of hours. I hacked a quick script to get tweets regarding the event. Using query #mitshooting, I collected more than 34 000 tweets in 10 hours of time. After only taking distinct tweets and cleaning up a lot of RT’s, spam and irrelevants, I had a little less than 6000 tweets containing over 40 000 words.
I’m a huge fan of wordclouds or tag clouds and first thing I wanted to do with the data was to create a wordcloud to see how which words were on the lips of twitter users.
The first cloud was created with wordcloud R library that I was familiar with from earlier clouds. It tells the story but is not my favourite when it comes to layout. With a little googling I found a service called Wordle and after uploading the same data there, I got something bit more beautiful with the same story.
For the changes in the tweets during the first 7 hours, I made a compilation
The code (which I’m still improving) can be found on my Github page