PoSh at SemEval-2023 Task 10: Explainable Detection of Online Sexism
Shruti Sriram, Padma Pooja Chandran, M R Shrijith · 2023
The growing popularity of online platforms and social media has intensified the problem of sexism, making it more prevalent and widespread.Online platforms have provided an anonymous space where individuals can freely express their misogynistic beliefs with discriminatory and abusive language, creating a culture of online harassment and hate speech, mainly targeting women.The dataset used in this task was provided by SemEval-2023 Task 10: Explainable Detection of Online Sexism.To precisely identify the different forms of online sexism, we utilize several sentence transformer models such as ALBERT, BERT, RoBERTa, DistilBERT, and XLNet.By combining the predictions from these models, we can generate a more comprehensive and improved result.Each transformer model is trained after pre-processing the data from the training dataset.Our team obtained macro f1 scores of 0.7937 for subtask A, 0.5284 for subtask B and 0.2674 for subtask C.