Yelp Reviews Help NYC Health Department Find and Close Dirty Restaurants 64
An anonymous reader writes with news about a study that investigated the effectiveness of Yelp reviews in pinpointing the source of foodborne illnesses. "In 2012, New York City's Department of Health and Mental Hygiene (DOHMH) found that residents weren't turning to the city's free 311 service to make such complaints, but rather they were reporting their experiences in Yelp reviews. So the CDC, in collaboration with the New York City DOHMH, Yelp, and Columbia University, conducted a nine-month long research into the effectiveness of using online reviews to identify sources of foodborne illnesses. The study discovered 468 actionable complaints, 97% of which hadn't been officially reported to the city, and analyzed roughly 294,000 Yelp restaurant reviews. Subsequent investigations on suspected restaurants turned up evidence of bare-handed food handling, cross-contamination, or even the presence of mice and cockroaches. The study concluded that providing the public with more options for reporting complaints about restaurants, particularly in the social media sphere, would help in the identification and possible closure of sources of foodborne illnesses."
CDC (Score:2, Interesting)
For others who, like me, did not know what the "CDC" was, it is "Centers for Disease Control and Prevention".
What would help? Doing their jobs (Score:2, Interesting)
Subsequent investigations on suspected restaurants turned up evidence of bare-handed food handling, cross-contamination, or even the presence of mice and cockroaches.
Subsequent investigations? That says to me that the initial investigation was much like a typical NYC building inspection. The "inspector" drives up to the business, glances around the front of the building, then tells you which pile of building materials they would like dropped off in their driveway before signing off.
Perhaps they should do their jobs which would result in finding things like mice and cockroaches, if not bare-handed food handling. Without that, my favorite solution to dirty restaurants (forcing them to post their health report in the front window, or in the same glass box as their menu) has no viability. Like public health, it depends on public health employees doing their jobs.
advantage is automation? (Score:2, Interesting)
I guess the big benefit here, is that very little human labor, on the part of the city, or the citizens, is required to find the suspicious restaurants. That suggests that having computers spy on people is more productive than having a web site to deal with customer complaints.
Improper methodology (Score:4, Interesting)
Those don't sound like serious violations, they sound like things you can find anywhere if you just look hard enough. I can see bare-handed handling and cross-contamination happening anywhere, and you'd pretty much need a hermetically sealed room to avoid mice and cockroaches in NYC.
How they should've done the study is mine Yelp for actionable complaints. Then send inspectors to those restaurants and an equal number of restaurants chosen at random without the inspectors knowing which set the restaurant belonged to. Then they could check to see if there was any statistical difference in inspection results between the Yelp-flagged set and the random set.
Otherwise you're just serving up a heaping of confirmation bias. The idea of using online reviews to detect food-borne outbreaks by mining review sites is a good one, but it still needs to be properly vetted in a double-blind study.