And don’t worry this isn’t going to be a rant about expenses.
At Moseley Barcamp I gave a presentation about ‘conversational psychogeography‘ and the potential for an explosion in data analysis of emotions and place. It was a very broad overview and I continually admitted that real research needed to be done (I’d love to do it, but might have to find someone to fund it). As part of the talk I mentioned by emotional analysis of Birmingham social media – which outputs at @birminghamuk on Twitter.
During the questions afterwards Tom Watson suggested that my Birmingham UK emotion scraping could be applied to groups of people â€“ Twittering MPs in particular. Technically it certainly can, and so I did. It’s at jonbounds.co.uk/arempshappy and tweets a score each day via @arempshappy on Twitter. It uses exactly the same code as the Birmingham UK system, except that it takes its data from an aggregated feed of the tweets of all MPs produced by Tweetminster.
Using groups of people immediately makes the analysis of the data easier (or at least offers precedent) â€“ there’s already a website offering (limited, automated) analysis of people’s tweets. It’s called tweetpsych and uses established linguistic techniques (eg LIWC) to produce scores for different aspects of personality based on person’s tweets. to extend this to groups of people should be easy.
Although restricting analysis to groups of people opens up possibilities, the groups need to be be large enough to make the sample useful and also static enough to keep the statistics baring comparison. In some respects the group of tweeting MPs is ideal â€“ and interesting enough to be worth analysing.
How useful it is is another matter, it was easy enough to set up â€“ and could be improved upon by someone interested in groups â€“ so let’s watch it and see if it spikes anywhere interestingly.