Typing out partial, unreadable tweets as you fall around looking for your Uber may be the go-to end a late night out in 2016. Nevertheless, those sloshed tweets that are might really be properly used for some good, big thanks to this machine-learning-based algorithm.
Several computer researchers from Rochester’s College is promoting a machine learning formula they’ve developed to identify tweets that were intoxicated.
More than 11,000 geotagged tweets were examined by the researchers, in New York City and Monroe County between July 2013 and July 2014. Out of this choice, they strained all of the articles that note alcohol-relevant buzzwords, including “drunk,” “tequila,” “beer,” “hammered,” and “get wasted”, giving various ideals and “weights” to every keyword, the computer can easily see if alcohol-drinking is clearly described, while remaining careful around deceptive phrases for example “shot,” “party,” or “club,” that might certainly not be about drinking.
By using this blend of drunken information and additional evaluation of words within tweets, the machine is subsequently ready to understand whether the article is about the Twitter-user themselves being drunk, and if that Twitter-user was really drinking at that time of tweeting. The computer was also capable to obtain the location of where the tweeter had been drinking whilst the tweets were geotagged.
We can analyze human mobility patterns; we can study the relationship between demographics, neighborhood structure and health conditions in different zip codes, thus understanding many aspects of urban life and environments. Research in these areas and alcohol consumption is mainly based on surveys and census, which are costly and often incur a delay that hamper real-time analysis and response. Our results demonstrate that tweets can provide powerful and fine-grained cues of activities going on in cities,” the researchers wrote.
The computer researchers expect that this algorithm may be used as a tool to deal with alcohol usage through public-policy, in addition to give a model for potential health-related surveys and censuses.