An Open Source API to help filter arabic text and check for toxicity and/or offensiveness
- dataset used Arabic-twitter-corpus-AJGT-
- inspired by every other bad word filter and the lack of ararbic one's (as far as i know)
- create basic model with +90% accuracy
- wrap around an api for ease of use
- add CORS
- jwt-auth
- rate-limiter
- security headers
- add a bad words filter
- update curses list to a txt/csv file
- fill the dictionary
- ingestion engine?
- /removeCurses --- bad word filter
- /getPrecentage --- AI
- /cleanCorpus --- remove punctuation and other chars