Three University Projects Use Twitter to Understand Happiness, Hate and Other Emotions in America

It turns out that the fleet­ing pro­nounce­ments we post on Twit­ter are cat­nip for aca­d­e­mics and oth­ers eager to find the elu­sive pulse of Amer­i­can soci­ety. Since Twit­ter launched in 2006, researchers have been hard at work fig­ur­ing out how to turn those 140-char­ac­ter mus­ings into tea leaves with some­thing mean­ing­ful to say about us all.

Here come three new projects that claim to pro­vide a win­dow into the Amer­i­can soul through Twit­ter. Whether they suc­ceed or not, well, that’s still unclear. (And, by the way, you can start fol­low­ing Open Cul­ture on Twit­ter here.)

Most fever­ish­ly excit­ed about its work are the team behind the Glob­al Twit­ter Heart­beat, which so far focus­es most­ly on the Unit­ed States. With the help of a huge SGI proces­sor to process a live feed of pub­lic social media data, a team of researchers from the Uni­ver­si­ty of Illi­nois at Urbana-Cham­paign has made a heat map to show how peo­ple react (through Twit­ter) to big events.

They looked at two things: Hur­ri­cane Sandy (top) and the 2012 Pres­i­den­tial Elec­tion (above). Using Twitter’s “gar­den hose feed”—a ran­dom sam­pling of 10 per­cent of the rough­ly 500 mil­lion tweets sent every day—researchers col­or-cod­ed tweets red for neg­a­tive tone and blue for pos­i­tive and showed the shift­ing con­cen­tra­tions of Twit­ter activ­i­ty across the coun­try. It looks like a map of a talk­ing weath­er sys­tem as occa­sion­al dia­logue box­es open up to show rep­re­sen­ta­tive tweets. Researcher Kalev Lee­taru argues that track­ing Twit­ter activ­i­ty gives us the poten­tial to track the heart­beat of soci­ety.

geographyofhate

Two oth­er projects look in an on-going way at tweet “tone,” or the negativity/positivity of mes­sages. One spin on this research is the Geo­graph­ic Hate Map (sam­ple map above), a project by Dr. Mon­i­ca Stephens of Hum­boldt State Uni­ver­si­ty in North­ern Cal­i­for­nia. To begin their work, Stephens and her team accessed a mas­sive data­base of geo­graph­i­cal­ly tagged tweets sent between June, 2012  and April, 2013.

They used only tweets that con­tained any of ten “hate words.” They read each tweet to be sure the words were used in a neg­a­tive way and built a map based on where the tweets came from. Then they aggre­gat­ed to the coun­ty lev­el and nor­mal­ized for the amount of twit­ter traf­fic in that area so that dense­ly pop­u­lat­ed areas don’t look more racist or homo­pho­bic by default.

Then there’s the glass half full. The Hedo­nome­ter mea­sures hap­pi­ness, or lack there­of, as expressed by tweets, cal­cu­lat­ing aver­ages based on what the researchers call “word shifts” (watch an expla­na­tion above). This research project, put togeth­er by the Uni­ver­si­ty of Ver­mont Com­plex Sys­tems Cen­ter, uses the same gar­den hose feed as the Glob­al Twit­ter Heart­beat. This project search­es for fre­quent­ly used words to mea­sure how good a day Twit­ter users are hav­ing. Since 2008 the Hedo­nome­ter has kept track of how often words like “hap­py,” “yes,” and “love” pop up in tweets, as opposed to “hate,” “no,” and “unhap­py.” The sad­dest day on Hedo­nome­ter record so far is April 15, 2013, the day bombs explod­ed at the Boston Marathon fin­ish line. Christ­mas Day tends to rank as the hap­pi­est day of the year.

To be sure, any tool that uses tweets for data is mea­sur­ing a very young and spe­cif­ic sub­group of peo­ple. Tweets are not a reli­able mea­sure of any­thing, real­ly, but maybe with some tweak­ing, these research mod­els will come up with some­thing inter­est­ing.

Relat­ed Con­tent:

The Art of Data Visu­al­iza­tion: How to Tell Com­plex Sto­ries Through Smart Design

Watch a Cool and Creepy Visu­al­iza­tion of U.S. Births & Deaths in Real-Time

An Ani­mat­ed Visu­al­iza­tion of Every Observed Mete­orite That Has Hit Earth Since 861 AD

Kate Rix writes about edu­ca­tion and dig­i­tal media. Fol­low dai­ly ups and downs on Twit­ter @mskaterix.


by | Permalink | Comments (0) |

Sup­port Open Cul­ture

We’re hop­ing to rely on our loy­al read­ers rather than errat­ic ads. To sup­port Open Cul­ture’s edu­ca­tion­al mis­sion, please con­sid­er mak­ing a dona­tion. We accept Pay­Pal, Ven­mo (@openculture), Patre­on and Cryp­to! Please find all options here. We thank you!


Leave a Reply

Quantcast