It’s estimated that 75 Million people use twitter. That may not sound like a lot when you consider that the rough population of our planet is about 6 Billion, but it is quite a bit if you were looking to indirectly poll a wide cross-section of people. Like perhaps, show the aggregate mood of a national population over the course of a day. Well that’s precisely what Alan Mislove, a computer scientist at Northeastern University in Boston, and a few of his colleagues have done.
They gathered their data sets from as strict a group as they can manage given the variables that can factor in to social networks. They only collected tweets from users in the US and mores specifically users who disclosed their location in their profile. This information is of course freely available to the public. The researchers then filtered their data based on something called Affective Norms for English Words, which is a psychological word-rating system. They then used the information they gained to produce a time-lapse visualization of the national mood as it progressed throughout the day. It’s really pretty interesting to look at, but what interests me more is the concept of polling for serious data from social networks. Imagine if we had a system setup to scan the stream of data from twitter for keywords like fever to help prevent outbreaks of disease. OR “ZOMBIE” to provide a bit of warning before the inevitable zomb-ocalypse! I don’t think that this sort of polling is terribly accurate tho. Or rather that it can only be so accurate as the filters the data is put through, though there are some variables I just can’t see being accounted for.
I can see some serious potential here if used correctly. Check the video below to see how your state really feels on the inside. (warm apple pie. ?)