Unmasking Hate: An Integrated Approach to Detecting Hate Speech in Social Media
Hate speech, a persistent challenge in human discourse, has found a fertile breeding ground in the digital era, posing a significant threat to our societal values and norms. The freedom of communication offered by online social media has given rise to an increasing prevalence of hate speech. This surge has resulted in cyber conflicts that impact social life at both individual and national levels. Consequently, detecting hate speech has aroused broad attention in the field of natural language processing and has been addressed in recent work. A recent method called dual contrastive learning has emerged, but it hasn't specifically focused on the transfer learning and emotion modeling aspects. Therefore, our research aims to leverage the capabilities of Dual Contrastive Learning (DCL) in conjunction with transfer learning and emotion modeling for the task of hate speech detection. Through this integration, our goal is to extend the existing dual contrastive learning framework for the detection of hate speech and analyze the contributions of transfer learning and emotion modeling to DCL's effectiveness in identifying hate speech.