Hacker News Reader

Comment by dynm | Hacker News Reader

I've always wondered why Amazon didn't take a "collaborative filtering" approach to reviews -- basically take your ratings and then give you recommendations based on others who tend to give similar ratings. It seems like this would mean that the scammers would mostly just give their nonsense reviews to other scammers.

View on HN

"Was this review helpful to you?" Maybe there should be a concerted effort of "people" to start clicking "No" to that question on obvious fake reviews. There's already sites tracking bad reviews, so "borrow" their list, create some bots, and start negating fake reviews. ??

Companies list/re-list products on amazon faster than you can say ‘go’ unfortunately.

This is one thing that I think YouTube gets right. Once a video has been posted and comments have been posted, liked, etc, you cannot change that video in any form. This is on the premise of avoid bait-n-switch to put of a video to garner the positive response and then swapping it out with a different video. Amazon should do something similar.

It's a lot harder. Google makes all the copies of a video. Amazon doesn't make all the copies of a product.

I don't thunk that would work with physical objects very well. Especially when recalls are done via serial number. Apparent identical physical objects are always fundamentally different from each other unlike data.

Just like movie recommendations. "Other people like you gave this 4 stars".

Combine that with a lot of machine learning and you could probably get a very accurate way to judge which products will delight which customers.

I wonder - could a third party do this? Amazon reviews could all be scraped...

It would be sucky in the same way "Hey you bought a new washing machine. Here is this new washing machine for sale!" recommended items often are.

Reply on news.ycombinator.com