You are viewing a single comment's thread:

RE: My Curation Desk: The Little Tool That Finds Real Life on Hive

It checks the curation history of top curators and uses that as part of its scoring. It also checks who the lower earners are and considers several other factors. So, it's already doing a kind of account health check, but I'll be adding more safeguards.

For OCD, it gives a higher score to users who have been curated by OCD this month, haven't received an OCD vote this week, and haven't been picked by me this month.

It also checks combinations of words used in the post and the tags to identify content that falls into the real-life category. The algorithm evolves with every yes and no.

KE ratio is also part of the scoring.

So, there's a lot happening in the backend, and the good thing is that I can keep improving it.

About open-sourcing the algorithm, @storicious asked about this too, and I was honestly too tired to reply at the time. The thing is, if I make my algorithm public, popular curators might start using it, and people could try to snipe my feed algorithm and find ways to break it. If I host it somewhere for my friends to use, I can't afford the hosting bills, and processing and learning from every curator's data would require a lot of computation.

TL;DR: I'm still at the beginner stage of this journey. I don't even know what's possible and what's not yet. I started working on one thing today and somehow ended up making three things, so I don't even know where the sky is or how to reach it. For now, I have no plans to open-source it. It's not that I don't want to share my work, but I don't want bad actors to exploit it either. We've seen in anti-abuse work how quickly people learn to get around things when they understand what's happening behind the black box.

-0.20453924 BEE
0 comments