Yelp reviews tend to drip with unrealistic expectations and ambitious prose, which is why the company recently decided to devote some of its resources to "coaching" its users to write more-thoughtful reviews.
But for all the icky entitlement and lofty haiku attempts that often take place on the restaurant ratings site, there have been exciting examples of case uses that serve a greater good. Like helping investigate outbreaks of gastrointestinal disease.
On Thursday, the Centers for Disease Control's Morbidity and Mortality Weekly Report detailed how Yelp reviews are being used by the New York City Department of Health and Mental Hygiene to find health code violations in New York City. It's the latest illustration of how big data and social media can be leveraged by resource-strapped city agencies.
Using software built by Columbia University, the agency combed through close to 300,000 restaurant reviews posted on Yelp between July 2012 and March 2013 to find possible health violations in the city, searching for cases of patrons reporting diarrhea, or vomiting after a meal. Of the 893 possible occurrences, 56% "described an event consistent with food borne illness," wrote researchers.
The data gives health code inspectors a window into the insulated kitchens of New York's sprawling restaurant biz, and allows them to send in the potential cavalry—read: health inspectors—if deemed necessary. All told, three New York restaurants were investigated as part of the experiment: One had cross-contamination between ingredients in the refrigerator and failed to properly sanitize its work surfaces. Another had bare-handed contact with ready-to-eat-food. And the third had evidence of live roaches and mice in kitchen, among other violations.
"With food-borne illnesses, it's much better to reach people sooner," Dr. Sharon Balter, a medical epidemiologist with the city, tells the New York Times. "When investigating an outbreak, we want to know what people who got sick ate, who else was with them and what items they all ate together. If you wait, people forget."
Like Google Flu Trends, one of the more contentious big data experiments, Yelp allows researchers to find potential unreported cases quickly without putting a significant strain on resources. The data is far from perfect, for obvious reasons. (If someone gets food poisoning, for example, there's no way to know if it was from something else they ate.) But it could help nab repeat offenders putting the public's health at risk.
The experiment is—perhaps thankfully—ongoing.