Last week, President Barack Obama kicked off his first Twitter town hall with – what else? – a tweet.
“At 1:53 PM Today (July 12), from the White House: I am going to make history here as the first President to live tweet,” he wrote.
And then he sent another tweet to get the conversation really going: “Today 2:07 PM Obama says: in order to reduce the deficit, what costs would you cut and what investments would you keep – both.”
Before it ended about an hour later, a number of well-known tweeters (e.g. House Speaker John Boehner) and lots of lesser known folks had tweeted hoping to catch the President’s attention and get a personal response to their question or comment.
While lots of commentary has already and will be written about this historic event, I thought I could provide something different (wearing my data scientist/data journalist hat) and parse his tweets, and those of previous town halls, if I could just recover the tweet steam. I used Searchtastic to retrieved 346 Tweets for visualization.
In the scatter plot (Chart 1 in slide gallery), one can see tweeters plotted by the number of followers versus the number of people they are following. The one stand out turns out to be Barack Obama in the upper right-hand corner with an out sized number of followers. Mousing over data points on a live version of the chart reveals additional details about who’s tweeting.
I was also asked to analyze the “most used words” and decided to start with the key tweets extracted in the AOL Live Blog and hacked them down to a spreadsheet and then plotted them using Spotfire software for analysis. The scatter plot (Chart 2 in slide gallery) shows a fairly random pattern of squares and colors suggesting very little repetition of words over the course of tweeting.
The tree map (Chart 3 in the slide gallery) alternatively shows the number of words by frequency (e.g. the word “to” was used 31 times). Interestingly, the word “economy” which was the essence of President Obama’s opening tweet: “in order to reduce the deficit,what costs would you cut and what investments would you keep” was only used three times!
There are certainly more powerful semantic analytics tools that the author is working with that could be applied to this kind of analysis.
These charts begin to parse the Town Hall tweets and wet one’s appetite for more of this kind of data science/data journalism. Its value has been discussed recently in Twitter Analysis As an Intelligence Tool in Libyan Engagement among other places.
I am working with Recorded Future to use their innovative tool to analyze tweets (Obama@TownHall, gov20, etc.) for their content and will report on that soon.