Sentiment Analysis of Hate Speech on Women in Social Media Platform Using Multi Label Classification

We live in a world where everything is connected to online social media platforms, and the person uses social media networks like Face book, Twitter, Instagram, Whatsapp, etc. In the present scenario, working women, celebrities, sports persons, and social living women are facing all kinds of comments and negative tweets on social media platforms. Even in many situations including online harassment, using abusive words, scolding toxic words, or hate speech increasing daily basis on social media platforms. In consideration of abused comments or words found on online forums, websites, blogs, YouTube, and social media networks must be detected using efficient automatic detection tools. Searching for each toxic word from comments is inefficient and requires a great deal of time. The study explored toxic comment datasets from Kaggle, using natural language processing; machine learning techniques with the DistilBERT model are used in sentiment analysis to identify the toxic text comments. Here DistilBERT model outperformed with the best accuracy of 97% results and machine learning technique with SVM using Multi classification obtained in Threat label with Precision of 0.99, Recall of 1.0, and F1Score of 0.99 and model accuracy is 0.99.