Drunk tweeters, beware: Computers are watching you.
A team of researchers led by Nabil Hossain at the University of Rochester have developed an algorithm that can pick up your tipsily tweeted messages with startling accuracy, according to the researchers.
Hossain and his colleagues had thousands of tweets analyzed through Amazon Mechanical Turk— a site that crowdsources human labor — to figure out which words and phrases best indicate drunkenness.
Once the robots figured out how to suss out human drunkenness, and differentiate from simply tweeting about drinking, the researchers came up with a map of what they call "unusual drinking zones."
By figuring out the ratio of geo-tagged tweets to drunk geo-tagged tweets in a given area, they managed to find the spots that had an outsized number of intoxicated Twitter users.
These are the unusual drinking zones of New York City, where your faithful reporter, her colleagues, and roughly 8.5 million other people reside:
It's pretty to see that Manhattan and Brooklyn have the highest proportions of tweets to drunk tweets. But those are big places, you say. What specific niches of the city should a teetotaler avoid?
The south end of Manhattan looks particularly tipsy. The population of the island almost doubles during work hours, so all those people grabbing after-work cocktails (and tweeting about it) add up.
Lower Manhattan is pretty much just a fog of presumed drunk tweeters, from Greenwich Village to Alphabet City.
As for Brooklyn, New York's reputed "hipster heartland," where the beer is craft and hats are floppy:
The population of some neighborhoods (though not all) skews young and upwardly mobile — a demographic set ripe for sending tweetpics of muddled cocktails with hand-chipped ice.
The neighborhoods that seem to have the most incidents of tweeting-under-the-influence are definitely renowned for their hipness, either established or "on the rise," as they say when suddenly dog salons start opening up on your block and rents skyrocket.
Trendy Williamsburg takes with the lead, with tony Park Slope and up-and-coming Greenpoint close behind.
It might be worth noting that these neighborhoods are actually quite different: Greenpoint is "in transition," which is often code for "gentrifying."
Park Slope, on the other hand, is known for being on the cutting edge of strollers and baby fashion.
Finally, let's take a look at Queens, because my editor lives there it is the most awesome borough in the city. Plus, I hear it's the new Brooklyn.
Apologies to the Bronx and Staten Island, but it seems like there aren't as many concentrated hotspots as we see in these three boroughs.
The paper's authors note that the areas that have a higher proportion of intoxicated Twitter users also tend to have a higher concentration of "alcohol outlets," meaning bars and liquor stores (shocking, I know). It also appears drunk-tweeting New Yorkers tend to drink at or closer to home than those upstate in Monroe County, New York, where the researchers also rounded up tweets.
How did they know where home was? By processing another set of geo-located tweets with phrases associated with being home ("home,""bath," etc.). Apparently they can use this to pinpoint a person's home within 300 feet up to 80% accuracy.
The authors hope to use the algorithm to eventually design an app that could intervene when someone's drinking heavily — maybe to flag their friends or others that could assist in extreme situations.
Join the conversation about this story »
NOW WATCH: Teens reveal their favorite apps and the winner is clear