The State of Profanity Obfuscation in Natural Language Processing