Identification and Labeling of Textual Cyberbullying using BiLSTM and BERT

In this aeon of digitalization, people are inclining more and more towards technology. This has both positive and negative aspects, and one of the negative effects is cyberbullying. Cyberbullying allows people to comment negatively on various social media platforms, this project aims at detecting and classifying such text. Using BiLSTM (Bidirectional Long Shortterm Memory) and BERT(Bidirectional Encoder Representations from Transformers), the cyberbullying text on various social media platforms is identified and classified into classes such as religion, age, gender, ethnicity, not_cyberbullying and other_cyberbullying. This helps in keep the record of types of cyberbullying that occurs on social media and whether it’s reduced over time after reporting it and taking strict measures against it.