Here are the frequencies for the spammiest tokens on the damus relay if anyone wants to use these for spam filtering:
Discussion
I feel sorry about this
Crap that’s a long list
Will do you consider using machine learning instead of key word to filter?
I can see so many Chinese words.😂
They may be testing the robustness of the nostr system.
Chinese is king of spam lol
This shows how #Nostr is substantially prized by the Chintoks, proof that "communism" is only a barrier to the emancipation of a ‘Chinese bitcoiner’ & that only #Nostr protocol could start a symbolic fight against the desire for communist collaboration leading only to the isolation of Asia-minor with regard to "decentralization & the possible adoption of Bitcoin"..
哈哈哈哈，用关键词来屏蔽吗？很快你们就会尝到火星文的厉害了。🐶
不得不说，国内这个spam和anti-spam行业太发达了，领先业界一百年。
垃圾人群。没灵魂，没思想。奴隶贱民只会搞这些垃圾事。
spammer是骂不死的👀
或者说，他们根本不看nostr。
That’s done on the relay side right?
#[0]
Is this similar with relay.nostrgraph.net , nos.lol , nostr.mom have found till now #[2] #[3] ?
I haven't checked the word frequencies. If anyone needs transparency I might do a bot in the long term that 'lightly' publishes the spam filter decisions as events on nos.lol. Obviously not every event rejected is going to be published about.. Fresh pubkeys will probably won't be mentioned at all.
there is no ends
These Chinese user groups.
It's really rubbish.
It's all a group of liars.
Cheating money.
Lose the face of the Chinese people. They are all low-quality garbage people.
Originally, Chinese people used American software to exchange ideas, do meaningful things, and do valuable things.
But it was ruined by this group of soulless garbage Chinese.
😤😡
#[0]
Can I filter spam from amethyst?
I noticed that the spammers basically have no followers, or less than 10.
We can use this feature to restrict people who post on public channels. If someone wants to post on a public channel, he must have more than 10 followers.
#[1] have you considered adding more structured ways of providing these feedback loops? For example ability to rate nip05 verifications and nip58 badges?
That way people can organically assess the trustworthiness of these forms of verification real time