Hacker News new | ask | show | jobs
Yelp Dataset Challenge (yelp.com)
26 points by y14 4332 days ago
4 comments

The prizes are ridiculously low.

Netflix did it right.

Stop being so greedy, Yelp.

Extremely low indeed. Most Kaggle competitions have higher prizes.

Btw, anybody knows if there is somewhere a list of those type of contests ?

Note that the prizes are only for students.
There are a lot of Kaggle competitions with similar prizes. Netflix is an outlier.
5k to help yelp? please.
Yes it's ridiculous, I'm just saying it's not atypical for these kinds of contests.
Is Yelp still penalizing listed businesses that don't pay them?
I often forget the exact name of the fallacy this type of question embodies, so for reference:

https://en.wikipedia.org/wiki/Loaded_question

The common analogue being, "Have you stopped beating your wife?"

That's not a loaded question when the person in question has a history of beating their wife.
This is why Yelp should kill two birds with one stone and just open source both their filtering algorithm and their rating algorithm. This would cause more engineers to want to work at Yelp and stop this PR fiasco where people think Yelp is gaming the system.

With the algo's out in the open we all can finally move on from this "extortion" rumor.

Do you have a source on that penalty? What kind of penalty is it, and does it relate to this challenge?
So sick of hearing this - what actual evidence do you have?

People's reviews get filtered because the user doesn't have human-identifiable data - multiple reviews, friends who use yelp, a picture, recent logins, etc.

Find me any business that complains about this extortion, their filtered reviews are 99% from people who've used yelp once and never logged in again.

On the other hand, here's a Harvard Business School study debunking this myth:

http://harvardmagazine.com/2011/10/hbs-study-finds-positive-...

What? That article has nothing to do with the subject of the myth.
Whoops, wrong link (on my phone, still no excuse).

Here is the one I was referring to: http://officialblog.yelp.com/2013/12/harvard-study-debunks-y...

The myth is that businesses which pay yelp do better; thus yelp extorts businesses to pay.

The study showed that businesses which pay for yelp ads do no better.

Thanks. But as skeptical as I am of the extortion claims, I don't see how this study debunks anything. All it does is show that there's no evidence of a systematic bias in the filter, but it doesn't disprove at all that Yelp manually tweaked results in isolated cases, which is what they were accused of.
5000$ is ridiculously low.
Wow that's a tiny carrot to dangle, Yelp.