Early working code (PoC) for Hate speech (Toxic content) detection
https://github.com/atrifat/hate-speech-detector-api
This API server will be included as part of dependency for nostr-filter-relay.
#machinelearning #deeplearning
Discussion
Detoxing speech in nostr?
It can helps to classify any potential hate speech (toxic) comments in notes (kind 1). It will gives prediction probability which can be used for filtering. Users can freely set their probabilty thresold (in percentage) as the filter settings on relay.
I have already implemented similar classification features in nostr-filter-relay which focus as "global feed" relay.
Update:
A public demo instance is available on HuggingFace Space - https://atrifat-hate-speech-detector-api-demo.hf.space. There is no guarantee for the uptime, but feel free to test this API server.
There is no Graphical UI (yet), so we can test it using tools like Postman, Reqbin, or Curl:
curl https://atrifat-hate-speech-detector-api-demo.hf.space/predict --header "Content-Type: application/json" --request POST --data '{"api_key":"", "q":"good morning"}'
Reqbin test tool: https://reqbin.com