Harnessing RNN for Enhanced Hate Speech Detection in Social Media

Social media offers a platform for both conversation and hate speech, making effective detection mechanisms necessary. This paper proposes a deep learning framework using RNN for hate speech detection on Twitter, Instagram, and Facebook. Key findings show that the RNN-based model outperforms LSTM and GRU models, achieving accuracies of 96.75% on Twitter, 95.3% on Instagram, and 98.20% on Facebook. These results demonstrate that RNNs better capture contextual relationships within text compared to traditional methods, while also emphasizing the need for specialized techniques in detecting hate speech across platforms. The pro- posed model holds significant potential for enhancing online safety through efficient hate speech recognition. The analysis confirms RNNs' superior accuracy for platform-specific hate speech detection, offering a powerful tool for improving detection strategies and online safety.