5 ms·

I've had similar ideas recently. Especially niche content (or shared research) would probably be notoriously hard (WRT false positives) for machine learning to

by eterps 7y ago

I've had similar ideas recently. Especially niche content (or shared research) would probably be notoriously hard (WRT false positives) for machine learning to decide whether it is relevant to you, people with similar interests know that much better. I was also wondering what would be good options to store votes/upvotes in a decentralized way.

worldsayshi 7y ago
> people with similar interests know that much better Yeah, I wonder if there is a cheap way to test this. Actually! There could be! Like using favorite's here on hacker news. That could be mined and visualized in various ways. (Although a quick sample shows me that it's a rarely used feature) > I was also wondering what would be good options to store votes/upvotes in a decentralized way. Yeah there are a lot of interesting optimization challenges if you really want to utilize upvote graphs for ranking.