Impact of Deep Learning Models On Hate Speech Detection
T Akhilesh Naidu, Shailender Kumar · 2021
Internet one of the mediums of connectivity that is available at the doorstep, with access to the internet one gets access to many web-based platforms. An increase in the use of these platforms gives us some benefits as well as some drawbacks. One of such drawbacks is hate speech. Hate speech is a topic of concern for social media platforms. With dynamically increasing datasets manual intervention of posts is quite impossible or will be time-consuming. Hate speech detection is an automated task to detect hate speech from the input. In this paper, we have compared some deep learning models like Convolution Neural Network (CNN), Recurrence Neural Network (RNN), Long Gated Recurrent Unit (GRU), and Long-Short Term Memory. The datasets used here are publicly available. The result of our analysis shows us that GRU performed better than other basic deep learning models. The model achieved an accuracy of 92.60% with an F1 score of 81.84% for dataset (D1) and the respective values for dataset (D2) are 96.15% and 83.06%.