We have noticed you are using an ad blocker
To continue providing news and award winning journalism, we rely on advertising revenue.
To continue reading, please turn off your ad blocker or whitelist us.
Although social media is part of most of our lives, many of us have been inflicted with a somewhat hilarious side effect, wherein messages have been posted when drunk. While, some may have participated or bore the brunt of drunken tweets, many have denied being under the influence. But now, scientists at the University of Rochester have developed an algorithm that identifies tweets that were posted when a user was intoxicated.
Researchers at the university collected over 11,000 geo-tagged tweets from New York and Monroe County between January and July 2014, which they then used as data to determine whether the person tweeting was simply talking about alcohol or was actually inebriated.
Given the amusing content of drunken tweets, they usually are not that difficult to identify for the layman, so one might wonder, why create an algorithm for it? One of the study's authors Nabil Hossain, along with the rest of the research team developed the machine learning algorithm, not just to detect drunken tweets but also to understand alcohol-related public health issues and its distribution across society.
Hossain and his colleagues isolated the tweets that were sent while intoxicated, from the data collected by filtering out all tweets that mentioned alcohol or alcohol-related activities, including containing words like "beer", "drunk", "party" and so on. The team then analysed the tweets using Amazon's Mechanical Turk workers. For each tweet, three Turkers were asked to determine whether or not a tweet was posted by an individual when drunk.
The research also focused on understanding where people preferred to drink (at home or a bar) and when and where they are most likely to tweet. The researchers intend the data to be applied in implementing better public health and alcohol policies in cities, based on the behavioural patterns revealed in the tweets.
According to the MIT Technology review, Hossain and his colleagues did not intend to just stop at analysing drunken tweets. Instead, they plan to study how consumption of alcohol varies with age, sex and ethnicity and how much peer pressure factors in as a contributor in influencing people to drink and tweet as well.
"We can explore the social network of drinkers to find out how social interactions and peer pressure in social media influence the tendency to reference drinking," the team of researchers said.
IBTimes UK has compiled a few (presumably) drunk tweets, check them out below: