Tracking Twitter Could Expose Restaurants Making People Sick

Looks like all those annoying Twitter and Foursquare users pulling out their phone in the midst of eating might actually be useful (full disclosure: we are also known to Tweet and Instagram our meals fairly often). University of Rochester researchers have devised a system, nEmesis, that compiles millions of Tweets to trace restaurant visits and coinciding food illnesses, tracking which restaurants are getting people sick.

For four months, the system collected 3.8 million tweets from 94,000 Tweeps in New York City; if the tweet is generated at a restaurant (combining the phone's GPS and the city's record of 24,904 restaurants), the system will track this person's tweets for 72 hours, scouring it for any mention of feeling ill. Of the 3.8 million tweets, the system found 23,000 at a restaurant, and 480 contained a report about food sickness (i.e. "stomach hurts," or "shouldn't have eaten that janky chicken. Gross").

The system then ranks the restaurants on how likely they are to get diners sick, and while individual tweets aren't always accurate, the researchers found that the rankings correlated with Department of Health grades. "The Twitter reports are not an exact indicator — any individual case could well be due to factors unrelated to the restaurant meal — but in aggregate the numbers are revealing," said Henry Kautz, co-author.

Foodborne illnesses are a major problem in the world, and have known to affect even the best restaurants in the world (re: Noma). So using this data to create timely reports on the quality of food at restaurants, in addition to the more intensive inspections from the DOH, could help users make "informed decisions," Kautz said.

Sure the system only takes in the experiences of people who tweet, but compiling Twitter data can give nEmesis users an extra layer of monitoring. A "seemingly random collection of online rants becomes an actionable alert," Kautz said. So keep oversharing, Tweeps.