Abstract:

Hate speech is a kind of communication that targets an individual or a group based on their race, ethnicity, religion, sexual orientation, or other characteristics. Although it may be conveyed in a variety of ways, both online and offline, the growing popularity of social media has expanded both its usage and intensity significantly. As a result, the goal of this study is to find and analyse unstructured data from chosen social media postings that attempt to promote hatred in the comment sections. To address this problem, we offer FADOHS, a new framework that combines data analysis and natural language processing methodologies to alert all social media providers to the prevalence of hatred on social media. We examine recent posts and comments on these sites using sentiment and emotion analysis algorithms. Posts suspected of containing dehumanising language will be screened before being sent into the clustering algorithm for further analysis. According to the experimental findings, the suggested FADOHS framework outperforms the state-of-the-art technique by around 10% in terms of accuracy, recall, and F1 scores.

Description:

Emotion recognition, clustering algorithm, sentiment analysis, data mining.