TWilBert for hate speech detection


The original work of TwilBERT is presented in the following paper: Gonzalez, Jose Angel, Lluís-F. Hurtado, and Ferran Pla. "TWilBert: Pre-trained deep bidirectional transformers for Spanish Twitter." Neurocomputing 426 (2021): 58-69. The license of the original work is BSD 4-Clause License.



TwilBERT is an adaptation of the BERT model trained with spanish tweets. This implementation has been finetuned for hate speech detection against immigrants and women in Twitter with the following dataset