Do you understand Yelp's filtering process?

In My Humble Opinion
1

I understand the need to distinguish reviews by people with an “agenda” from other more legitimate reviews - no question. But I’ve watched their video explanation twice now, and it doesn’t seem to do a very good job of explaining the process (perhaps it’s in their own best interest to keep it ambiguous).

I’ve used Yelp off and on over the last six months, but have yet to write a review on any local businesses. It just seems that some of the reviews that get filtered (and therefore probably unread by what? 20%? 50%? 90%? of the population) seem to be legitimate.

Any ideas what their reasoning is? Have you written any reviews that did/did not get filtered? Perhaps it’s a case where if you have less than 10 reviews, they’re all considered “suspicious” until you establish that you don’t have an agenda, and aren’t just a typical internet troll?

2

I strongly suspect they are using machine learning algorithms to classify reviews as legit or not-legit. The algorithms look at everything from sentence structure to pronoun usage to the words used. They train the algorithms by having them look at tons of reviews that are pre-labeled as real or fake, pick up the patterns in them, and then apply those patterns to new reviews.

3

I think that the number of reviews you write is taken into account. My first few were filtered but now I never am.

4

I’ve written several reviews and had them filtered out. The result being that I don’t bother with Yelp since I know legitimate reviews get taken out wholesale.