Re:Reddit overhauls upvote algorithm to thwart cheaters
by minimaxir on Wednesday December 07, 2016 @12:57PM
(#13124534)
As a proof of concept, I made a simple related subreddit finder using solely the commenting data of users: http://minimaxir.com/2016/06/reddit-related-subreddits/
The "like X subreddit and Y subreddit? Here's Z" is not infeasable to do with ALS on a similar dataset, but requires a better computer than I have. :p
Re:Reddit overhauls upvote algorithm to thwart cheaters
by EvilTerran on Wednesday December 07, 2016 @01:08PM
(#13124659)
So something along the lines of... a new tab on the homepage, "You Might Like: posts upvoted by people who tend to vote like you"?
That is: compute a "similarity score" between each pair of users, higher for each post they both voted on the same way, and lower for when they voted oppositely; then, when computing the ordering for the "You Might Like" page, multiply each vote on the posts by the voter's similarity to the viewer, instead of counting each voter equally.
Or something like that - I'm just a layman tossing ideas around, no doubt there's more rigorous and/or performant statistical methods out there.
I think there'd be a lot of interesting discovery available to reasonably active users of reddit. (I've spent a significant % of my career on personalized recommendations, and in my opinion reddit's product is close to ideal in many regards).