Ask HN: Distributed content ranking learning resources?

3 pointsby cryptarch10 jaar geleden1 comments

I want to get my hands dirty with distributed systems, and my main interest lies in building distributed alternatives to authority- and identity-based voting systems for content, to make it possible to have distributed open-data projects. I want content curation without centralized authority.

I'm looking for a way to curate content with the following properties:

* Decentralized ranking

* No centralized identity provider

* Resistant to vote manipulation

* Optional manual training for better filtering/ranking

* Optional (partial) trust for 3rd parties that have aggregated manual training data (like Adblock filter lists) or explicit rules

I would want to use this to build alternatives to forum software like Reddit, and also to make open data projects containing i.e. cooking recipes or media metadata easier to decentralize.

I'm thinking some kind of web-of-trust implementation paired with content/behaviour analysis to segregate human users from bots (ideally to the point where bots would have to post actually "good" content to compete), but I'm completely stumped as to where to even start researching.

I have seen Aether, a decentralized Reddit alternative, but it's seems to be inactive until the next release.

What techniques and/or papers would you recommend I start with? Do you know any good book or tutorials on this topic?

Edit: wording, made title clearer

1 comments

brudgers10 jaar geleden

In terms of techniques, I would suggest writing some spike code to flesh out ideas.

Good luck.