Scientists told how to identify a drunk person on Twitter

Researchers from Rochester, NY has developed a program to automatically detect people who use alcohol. As the primary source of information used messages from the social network Twitter.

All scientists have collected 11 thousand tweets selected using the filter word: "beer", "party", "bar", "drunk". Messages proposed to analyze specially hired people. So the volunteers have divided the publication into two groups: discussion of alcohol or drinking at the time of writing.

But the basis of the obtained data, the researchers created a program for automatically determining the "drunk tweets". Now, using established tools and geolocation will be possible to make a list of the most popular places for alcohol consumption.

Subscribe to new posts: