Twitter’s Hate Speech Multi-label Classification Using Bidirectional Long Short-term Memory (BiLSTM) Method

Abstract

Since social media is one of the most likable products of techngetpeople get easier to express their opinions. Anyone be able to tell their opinion freely there. Unfortunately, its convenience has also become a boomerang for us, the easier every opinion conveyed the easier hate speech is expressed. This matter become the dark side of social media. Hate speech face us with a lot of dangers, such as violence, social conflict, even homicide. Therefore, preventing all of those dangers that might be occur because of hate speech is one of the prior things we need to do. This research was done as an attempt to take care of the dangers that could be done by hate speech. The attempt we tried to do is using multi-label text classification to predict hate speech with the Bidirectional Long Short-term Memory (BiLSTM) method. This multi-label text classification labelled every tweet in the dataset crawled from Twitter with 12 labels about hate speech. From this experiment, we obtained the best hyperparameter value that could achieve great performance with 82.31% accuracy, 83.41% precision, 87.28% recall, and 85.30% F1-score.