Hi!
I’m working on reputation systems and have experimented with the !praise data.
Most reputation systems fail due to not being contextual enough – reputation is multi-dimensional. Universal scalars tend to become social credit scores, due to the perceived objectivity of such numbers.
I’ve built a system that uses language embeddings and PR to determine who is most reputable within a certain area in the DAO based on the !praise data.
You can try it like this:
curl -X POST -H \
"Content-Type: application/json" \
-d '{"query": "graphic design", "server": "tec", "api_key": "c4a0ae09-4dab-41d3-9660-d74d551f44c5"}' \
https://replabs-flask-app-aucndxjanq-ew.a.run.app/query | jq .
The “query” parameter is dynamic and can be any string representing the type of reputation you’re interested in. For example, “graphic design”.
I mainly did this as an early experiment for https://replabs.xyz/, but thought I’d see if this sparks any interest in the TEC community – do you see this being useful to you?