Ensemble Deep Model for Hate Speech Detection

Over the last decade, online platforms such as Meta and Twitter have experienced a significant increase in data due to a surge in users. Unfortunately, this growth led to an increase in hate speech content on social media platforms. The detection of hate speech is critical for various applications, including sentiment analysis and abusive content identification. Deep learning models have become increasingly popular for detecting hate speech online. This study employed three deep learning models, namely Convolutional Neural Network (CNN), Bidirectional Long Short-Term Memory (Bi-LSTM), and Robustly BeRT pre-training approach (RoBeRTa) model for hate speech detection. Additionally, we proposed an integrated model that combined CNN, Bi-LSTM, and RoBeRTa and evaluated the models using a dataset of hate speech extracted from Twitter. The integrated model produced better results than the other models evaluated.