The death of Michael Jackson, the Japanese Tsunami in 2011, and the 2012 school shooting in Newtown, Conn. - the dates of these events are registered as some of the saddest days on Twitter since 2008, according to a new study led by the Computational Story Lab at the University of Vermont Complex Systems Center.
The project is on display as an online, interactive graph titled the Hedonometer, or, as researcher Chris Danforth has called it, "the Dow Jones index of happiness." The project was developed with the technology of Brian Tivnan, Matt McMahon and their team from The MITRE Corporation.
"Happiness is difficult to measure, but we've built the prototype of an instrument that is sensible and improvable," Danforth told ABC News.
"You can think of this instrument as operating in the manner of a thermometer," he explained. "Each molecule represents one tweet, and in aggregating hundreds of millions of words we get a sense for the emotional temperature of large populations."
The project looks at a "random sampling" of roughly 50 million tweets daily, or about 10 percent of all tweets posted each day. Researchers created a set of about 10,000 keywords to be detected by the system. Each keyword is given a score from sad (1) to happy (9), and an average happiness score is then calculated for each date.
So, where do the peaks and valleys of happy tweets lie on this chart?
The Hedonometer shows Christmas and Christmas Eve as being the happiest days of every year included in the study, with keywords like "merry" and "happy" in high occurrence. Other holidays like Valentine's Day, Easter, Mothers Day and 4th of July each show a spike in joyous tweets including the words "happy," "love" and "family."
As for the not so happy times on Twitter, those dates and keywords seem to correlate with some of the most tragic events of the last five years. The single largest dip in happiness was on April 15, the date of the bombings at the Boston Marathon that left three dead and over 200 injured. On that day, Tweets containing the terms "explosion," "sad," "prayers" and "tragedy" crowded the Twitter-verse.
Other sad dates, according to the Hedonometer: December 14, 2012 (the day of the Sandy Hook Elementary shooting in Newtown, Conn.) and September 29, 2008 (the day a U.S. bailout bill meant to address the recent market crash failed in the House of Representatives).
The graph also shows May 2, 2011, the day the United States announced Osama Bin Laden had been killed, as one the saddest days in the study.
"It's clearly a complicated day," said Danforth. "Regarding public opinion on the event, I think Twitter is one natural place to look for information about how people were feeling that day.
"Many people presume this day will be one of clear positivity. While we do see positive words such as 'celebration' appearing, the overall language of the day on Twitter reflected that a very negatively viewed character met a very negative end," he explained.
"We get news quickly, especially when tragic events happen," Bryan Reuther, postdoctoral resident in clinical psychology at Nova Southeastern University told ABC News. "Social media platforms such as Twitter provide people with an immediate outlet which, in many ways, is known to reach more people in a shorter time."
Reuther believes that people turning to online social networks to express feelings is becoming more of a new normal.
"Emotional expression is key to being human and emotional expression over social media is becoming part of the 'plugged in' human," said Reuther.
Image credit: Christopher M. Danforth/University of Vermont
So does the overall dip in the Hedonometer from 2008 to present mean the level of sadness amongst tweeters has increased?
Danforth said, "The demographics have certainly changed during this time period, so some of the decline is a result of the shift in population. We do see simple negative words (e.g. 'no', 'never', "won't") being used more frequently now than a few years ago."
As for the big picture aim for the project, Danforth said his team would really like to provide an alternative measure of population level well-being to complement more traditional measures like GDP and Consumer Confidence.
"Our next improvements will be to include a dozen other languages, as well as information about what people are searching for (Google Trends), reading (traditional news media), and sharing (bitly) online."