A Transformer Based Approach for Abuse Detection in Code Mixed Indic Languages.

Computer Science

• web user 100%
• Machine Learning Model 100%
• Language Modeling 50%
• Bidirectional Encoder Representations From Transformers 50%
• Textual Content 50%
• Individual Model 50%
• Handling Mechanism 50%
• Cyberbullying 50%
• Low-resource Languages 50%
• Social Media 50%

Keyphrases

• Transformer-based Approach 100%
• Classical Machine Learning 66%
• ROC-AUC 33%
• Four-state 33%
• Bidirectional Gated Recurrent Unit (BiGRU) 33%
• Multilingual BERT (mBERT) 33%
• Emoji Embedding 33%
• Code-mixed Language 33%