And now for a brief St. Patrick’s Day PSA: If you’re planning any celebratory outings today, drink responsibly and do your level best to not drunk tweet or text. It may seem like a good idea at the time, but the Internet is forever.
Thanks to context clues, you’ll likely be able figure out if you’re on the receiving end of one of these inebriated messages, and now a computer does too, thanks to some researchers at the University of Rochester. They recently built an algorithm to identify tweets sent under the influence and observe general “alcohol-related activity.”
The algorithm that head researcher Nabil Hossain and his team developed can sort out tweets about alcohol and drinking and tweets sent by people while they are imbibing.
The project was a couple of years in the making, with about 11,000 geotagged tweets collected during 2014 that were sent from New York City and Monroe County, N.Y., (where the university is located) to better figure out where these individuals were tweeting from, such as at home or at the bar, down to about 100 meters with 80 percent accuracy.
Going forward, the computer scientists want to take what they’ve learned and examine how a variety of venues — such as a stadium on game day or a friend’s house party — affect drinking and tweeting and how alcohol consumption differs among age groups, ethnicities and genders.