<-- Advertise here.

The scientific study "Pulse of the Nation: U.S. Mood throughout the Day Inferred from Twitter" [ccs.neu.edu] illustrates the varying mood in the U.S., as inferred after analyzing over 300 million tweets that were created over the course of the day. Various density-preserving cartograms and a time-animated video were produced to capture important large-scale trends.

The researchers analyzed all public tweets posted between September 2006 and August 2009, and filtered those whom orginated from a US location and those that contained words included in the psychological word-rating system called Affective Norms for English Words, resulting in a collection of 300 million tweets. Through a natural language processing algorithm called Sentiment Analysis, each tweet was assigned a mood score based on the number of positive or negative words it contained. Out of the resulting data, they then calculated the average mood score of all the users living in a state hour by hour which formed the basis of a series of time-varying mood maps.

One of the interesting patterns shows how the West Coast mood follows the same pattern as the East, with a 3-hour time-zone delay, indicating that Each Coast experiences the same time-dependent swings. Weekends were observed to be happier than weekdays. The peak in the overall tweet mood score is observed on Sunday mornings, and the trough occurs on Thursday evenings.

Get the full paper here. Via Newscientist and Slashdot.



This is really well done! I love it. Very interesting data perfectly visualized. Using cartograms with mapping the mood to colour and number of tweets to region size reveals an unexpected pattern (at least for me).

Fri 23 Jul 2010 at 9:30 PM

Great to see experimentation like this but I don't think the finished visualisation actually provides a particularly useful means for people to explore the data or draw insight. I've compiled some potential alternative design approaches here http://www.visualisingdata.com/index.php/2010/07/twitter-visualisation-of-happiness/

Fri 23 Jul 2010 at 11:00 PM

There appears to be an inverse relationship between the US and the EU. Interesting...

Sat 24 Jul 2010 at 9:24 AM

This sort of thing is more fun than scientific. Twitter is a very unrepresentative sample of the US. In fact, only about 7% of people who are online are on Twitter, and they are a unique group of people.

Regardless, I would love to see whether this result holds for conversations that take place elsewhere, such as on YouTube, Blogger, FaceBook, etcetera. Replicability across divergent sources will reveal the true validity of these findings.

Annie Pettit
Conversition Strategies

Wed 28 Jul 2010 at 1:23 AM

What I take away is this: people who are on Twitter in Florida and California bitch less that the rest of the country.

Wed 28 Jul 2010 at 1:33 AM

Hello, there is a mistake in the current post. The work you're writing about, "Pulse of the Nation: U.S. Mood Throughout the Day inferred from Twitter," is by: Alan Mislove, Sune Lehmann, Yong-Yeol Ahn, Jukka-Pekka Onnela, J. Niels Rosenquist.

However, at the bottom of the post, you say "get the full paper", but link to a different study by myself (Brendan O'Connor), Ramnath Balasubramanyan, Bryan R. Routledge, and Noah A. Smith, entitled "From Tweets to Polls: Linking Text Sentiment to Public Opinion Time Series."

We did not do the study that appears in the poster that this blog post and the linked articles are talking about! Please give its authors proper credit. I guess these two studies have some similarities, but this is a serious mistake in attribution.

I don't want people accusing me of stealing credit for someone else's work, and this blog post has already led to some confusion.

Wed 11 Aug 2010 at 3:57 PM
Commenting has been temporarily disabled.