A new system has been developed by scientists at the University of Rochester in New York which analyses tweets about restaurant visits.
The nEmesis system ranks restaurants in order of food safety by indicating how likely it is for someone to become ill after eating out. It could enable what food inspectors call "adaptive inspections," inspections guided in part by the real-time information that nEmesis provides.
The system combines machine-learning and crowdsourcing techniques to analyse millions of tweets to find people reporting food poisoning symptoms following a restaurant visit, which would be impossible to analyse manually.
The technology was first launched in New York City four months ago. In that time, the system collected 3.8 million tweets from more than 94,000 unique users in New York City, traced 23,000 restaurant visitors, and found 480 reports of likely food poisoning. They also found the tweets correlate fairly well with public inspection data by the local health department.
"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, chair of the computer science department at the University of Rochester and co-author of the paper.
nEmesis can identify the restaurants by combining the data of where the person tweets from, usually from a phone, and the known locations of restaurants. This means that tweets can be "geotagged".
If a user tweets from a location that is determined to be a restaurant, the system will continue to track this person's tweets for 72 hours. If a user then tweets about feeling ill, the system captures the information that this person is now ill and had visited a specific restaurant.
Vincent Silenzio, co-author of the study, said: "People criticise folks for over sharing on Twitter and social media, but there's a benefit.
"Currently you can only identify a [bad] place after the fact. The promise of this system is that you can identify things almost in real time and provide a better detection system than you have now."