Opinion Mining for the detection of Hate Speech on Social Media

Abstract

Since hate speech and offensive language is becoming a growing problem, with the massive availability and popularity of opinion-rich resources such as social media, online review sites Hate speech on social media has unfortunately become a common occurrence largely due to advances in mobile computing and the internet. This thesis work analyses the social media content to try and identify hate speech, and with the application of various machine intelligence models, improving the detection rate with higher accuracy in contrast to previous research works. Various deep learning approaches such as Convolutional Neural Network (CNN) and Recurrent Neural Network (RNN) are used in the proposed approach. With the application of sentiment weighing, the opinion mining is carried out. The final model shows overall precision of 0.82, recall 0.78 and F1 score of 0.80. The test accuracy achieved in the dataset is 83%. For the visual interpretation, ROC curve shows the distinction between the abusive or offensive and non-offensive content.