A Team of Grad Students Has Developed a Tool to Root Out Fake Hotel and Restaurant Reviews

Not simply The New York Times, but a number of other worthy publications, have recently written about the effort of three graduate students at Cornell University to create a computer program that would enable user-generated websites to delete fake reviews. Yet in explaining the formulas and algorithms in their program, those students have also given tricksters a sure method for avoiding being caught.

A key factor in spotting a fake hotel and restaurant review, according to the Cornell students, is the absence in the fake recommendation of any physical characteristics of the hotel or restaurant. That's because the author of the fake recommendation has never been to the establishment in question.

But starting now, all that needs to be done is the insertion of a statement referring to "pleasant cream-colored walls of the coffee shop." That, in the logic of the new computer program, will immediately mark the review as authentic. And if the writer wants to be especially sure that his or her recommendation will pass muster, they need only pull up a photograph of the establishment on the internet, and comment about the actual physical characteristics of the establishment.

Other aspects of the new algorithm will also provide the secret of avoiding being caught. It's obvious that in the future, skillful writers of recommendations, or skillful authors of critical condemnations, will easily be able to avoid whatever patterns have been discovered by the Cornell whiz kids. Although these graduate students are now being offered jobs by the frantic, worried, recommendation sites (who all realize how serious is their inability to filter out fake reviews), the effort seems bound to fail.

To repeat my mantra: if a hotel is dependant on TripAdvisor and the like for its success, why won't it generate its own reviews? And how difficult is it to write a fake review that's indistinguishable from a genuine review?