Working on a generic, deterministic scoring engine for ts-profanity-filter. It combines weighted signals, modifiers, patterns, token distance and confidence into an explainable result.

I'm looking for contributors interested in algorithms, NLP, TypeScript or content moderation. Feedback and ideas are also very welcome.