- How to watch ‘Game of Thrones’ season 8, episode 2 for free Today 7:00 AM
- Gendry is making a new weapon for Arya Stark—but what is it? Today 6:30 AM
- The live-action Halo series could be Showtime’s most ambitious project yet Today 6:00 AM
- How to watch Turner Classic Movies for free Today 5:30 AM
- How to watch Real Madrid vs. Athletic Bilbao online for free Today 5:00 AM
- ‘Star Trek’s Jonathan Frakes calls out your lies with this new meme Saturday 3:46 PM
- #JusticeForLucca trends after video shows police slam Black teen’s head into pavement Saturday 3:11 PM
- The internet is shocked to learn that Goombas do, in fact, have arms Saturday 2:02 PM
- PayPal, GoFundMe cut off armed militia that detains migrants at border Saturday 1:16 PM
- Barnwood theft may be on the rise because of ‘Fixer Upper’—and fans aren’t having it Saturday 12:23 PM
- Literary Twitter calls out Dzanc Books for Islamophobic, racist novel Saturday 11:40 AM
- How to watch Crawford vs. Khan online Saturday 10:00 AM
- Beyoncé has 2 more projects coming to Netflix after ‘Homecoming’ Saturday 9:53 AM
- How to watch Danny Garcia vs. Adrian Granados for free Saturday 9:00 AM
- The ‘Feeling Cute Challenge’ turns ugly after correctional officers abuse it Saturday 7:30 AM
It may help with public health.
Sometimes, when you’re drunk, you want the whole world to know. So maybe you take to Twitter.
vodka is tricky, cause 1 second you think you’re a supreme heavyweight and its not effecting you, & the next you’re chatting up a chair
— drunk (@drunk) March 11, 2016
But now scientists have developed a machine that can not only learn where you’re drinking, but whether or not you’re drinking locale is at home or at a bar. According to the MIT Technology Review, the work could have “significant impact on the way we understand and respond to the public health issues that alcohol and other activities raise.” In particular, the research raises questions about whether people drink more than they normally would if they live in an area with a lot of bars, or whether people who drink more tend to prefer to live in these areas.
The researchers rounded up a bunch of tweets geotagged in New York City and the state’s Monroe County and filtered all the tweets that contained words like “drunk,” “party,” “beer,” and more. But before they could teach the machine to determine if these tweets were the tweeter referencing drinking at that time, or drinking in general, the researchers used humans sourced through Amazon’s Mechanical Turk to figure it out first.
The researchers also wanted to know if tweeters were tweeting from their homes or elsewhere. So they devised a number of different ways to determine that, such as looking at the geotag of the tweet prior to the “drunk tweet,” where they typically send their final tweet of the day from, and where they tend to tweet between the hours of 1am and 6am. They also had the Mechanical Turk volunteers determine if certain tweets that contained references to household items, like a bath, were sent from the tweeter’s home.
Using this data, they made an algorithm that would allow a machine to learn a tweeter’s pattern and identify alcohol-related tweets.
According to the MIT Technology Review, the algorithm can pick out a person’s home within 100 meters and with 80 percent accuracy.
Understanding how, when, and why people drink—as well as who’s doing the drinking—can help researchers understand factors related to public health issues like alcoholism and drunk driving. Getting that data from Twitter happens to have the benefit of being cheap and easy, compared to going out into the field and conducting lengthy surveys to answer the same questions.
However, it may make you think twice before pulling your phone out when you’re bored at that house party.
H/T MIT Technology Review | Illustration via Max Fleishman
Cynthia McKelvey covered the health and science for the Daily Dot until 2017. She earned a graduate degree in science communication from the University of California Santa Cruz in 2014. Her work has appeared in Gizmodo, Scientific American Mind, and Mic.com.