Classification of Multi-Label of Hate Speech on Twitter Indonesia using LSTM and BiLSTM Method

Aurora Az Zahra, Elita
Sibaroni, Yuliant
Suryani Prasetyowati, Sri

JINAV: Journal of Information and Visualization
Vol. 4 No. 2 (2023)

Social media is a communication tool that supports users to interact socially using technology. One of the most popular social media platforms is Twitter. However, its media platform has been considered by the virtual police as one of the main sources of spreading hate speech on social media. In this final project research, the authors conducted a study on the detection of hate speech in tweets on Twitter Indonesia. The method used in this research is multi-label classification by applying the LSTM and BiLSTM methods. The dataset used was 13,169 tweet data, and data labeling process was carried out into 12 classes. The results revealed that the LSTM and BiLSTM methods had good performance in classifying text data with 10 trials with an accuracy value of 78.67% for LSTM and 80.25% for BiLSTM. Based on the accuracy obtained, BiLSTM has higher accuracy than LSTM, so it can be concluded that BiLSTM is superior to LSTM.