@banneret @pandora_parrot Almost like a half-step between mute and block. It would be hard to have effective CW text though..
@banneret Fuzzy matching allows a user to pick the words to use, which could help.
@Doephin
Definitely better for individual users, whereas text categorization is more of an instance level solution. I think there are eighty, ninety words I would put in my personal dictionary, maybe a dozen bigrams as well?
@Doephin
The main issue is selecting the right words. We could categorize existing toots and use a supervised learning solution to categorize incoming text. Alternatively, just fuzzy match on a manually generated dictionary?
@pandora_parrot