A combined human-AI approach to spam labelling
A threshold number of spam reports triggers an AI bot
The bot reviews a self-selected sample of notes by the flagged account
The bot publishes a timestamped "spam" or "not spam" label
No further evaluations of that account are permitted until X time has passed
Clients can use "spam" labels to filter accounts outside of the user's social circle (follows & follows-of-follows, WoT etc)
Censorship-resistant and mistake-resistant
Can be applied to other types of content labelling
tf
npub1dq…8yffq
2024-07-15 06:34:46
Author Public Key
npub1dq0vnsph9wpmp9pcsc6qj2l5h7xsjyxyxt5tsl986u3v4lncknnsf8yffqPublished at
2024-07-15 06:34:46Event JSON
{
"id": "e66762b6508cfb83a498a2f95ba80ce7c84b3df3390a7c748d42921f26882f6c",
"pubkey": "681ec9c0372b83b094388634092bf4bf8d0910c432e8b87ca7d722cafe78b4e7",
"created_at": 1721018086,
"kind": 1,
"tags": [],
"content": "A combined human-AI approach to spam labelling\n\nA threshold number of spam reports triggers an AI bot\n\nThe bot reviews a self-selected sample of notes by the flagged account\n\nThe bot publishes a timestamped \"spam\" or \"not spam\" label\n\nNo further evaluations of that account are permitted until X time has passed\n\nClients can use \"spam\" labels to filter accounts outside of the user's social circle (follows \u0026 follows-of-follows, WoT etc)\n\nCensorship-resistant and mistake-resistant\n\nCan be applied to other types of content labelling\n\n",
"sig": "680b2f031c04aba176008675eb89febee2f7080ab23791fb040499ede61d72f5aff8364d5ce75f802948c98d00a350c444e725be2dbaf3ca6a9b5d97532934d7"
}