Skip to main content
. 2020 Aug 27;15(8):e0237861. doi: 10.1371/journal.pone.0237861

Fig 1. The proposed framework for hate speech detection and bias mitigation tasks.

Fig 1

It consists of two different modules: Hate Speech Detection and Bias Mitigation with different inputs as a result of different pre-processing approaches. The pre-trained BERTbase is a common component between two modules that is fine-tuned differently in respect of each module’s goal.