Looks like a new academic discipline is born-social media sociological forensics. Say what? Academics used Twitter tweets to find a correlation between the amount of anger online and the anger on the streets of London, which blew up into looting and violence. Looking through the general mood of tweets, scientists from Bristol University found that the level of fear and anger rose during 2011. According to their report, presented at a Lyons, France international workshop, the amount of anger also dipped right before the Prince William and Kate Middleton’s Royal Wedding. This suggests that the royal wedding had a calming effect on the populace. This was no small or superficial survey. The research pored through the almost 500 million 140 character or less microblog public posts sent by 10 million users. The tweets were sent between summer 2009 and the early part of 2012. The study aimed to see the UK’s general mood during this period of pronounced economic downturn.
The public mood reacted differently to different government pronouncements. For example, negative sentiment rose when the UK government announced cuts to public spending and other austerity measures. According to Nello Cristianini, co-author of the research, social media enables researchers to easily gather heavy volumes of public communication and allow for correlations between events and social media mood trends. Still, he did not close the door to the possibility that other events might have caused the spike in social media moods.
At this point of the story, you and other intelligent readers might probably be wondering how the researchers were able to assign the 500 million tweets (imagine that volume!) to specific moods and public sentiment. There are quite a wide range of human emotions and the huge number of tweets can easily fall between such a wide range. Moreover, written messages on the Internet don’t easily convey emotions in a standardized way like tone of voice, facial expressions, and demeanor in face to face communication.
Despite the limitations above, the researchers’ analysis of tweets involved running the tweets through a list of filter words, which were linked to different feelings and moods. While we understand this is the most efficient way to handle such a heavy volume of tweets, this might not be the best way due to the emotional ambiguity of many typed communications. Also, there were no contextual factors put in the filtration process.
Still, the research does indicate a pattern emerging where key times during the year like Halloween, Valentine’s Day and Christmas producing certain mood patterns.