Reminds me of Jonathan Harris's 2007 project 'We Feel Fine.' http://wefeelfine.org/methodology.html http://wefeelfine.org/methodology.html When I did some wor

Reminds me of Jonathan Harris's 2007 project 'We Feel Fine.' http://wefeelfine.org/methodology.html http://wefeelfine.org/methodology.html When I did some work on sentiment analysis using tools like Crimson Hexagon, it always felt like the data was skewed since what people post is different from what they say, which itself is different from what they do. Might need to be some kind of corrective filter that includes 'uncounted emotion' as a baseline.