The Ranking of hate speech propagators in social media using a multimodal profiling framework

Abstract

Social media platforms have revolutionized communication by enabling the sharing of opinions, ideas, and beliefs at an unprecedented scale. However, the rapid growth of these platforms has also facilitated the widespread propagation of hate speech and harmful behaviour, posing both technological and societal challenges. Understanding hate speech propagators, their influence, reach, and impact is crucial for mitigating the spread of harmful content and fostering safer online communities. This research aims to address this challenge by developing a robust hate speaker ranking system to identify, analyze, and rank hate users based on their behaviour and influence. The proposed framework combines profile features, characteristics, and machine learning techniques to evaluate hate speech. The system provides a dynamic, multimodal ranking mechanism to rank hate users into three levels labelled as: HIGH, MEDIUM, and LOW, based on their level of influence. The framework is trained and validated on datasets from Twitter, ensuring scalability and adaptability across diverse environments. Results indicate that the system effectively identifies influential hate users with high accuracy, offering actionable insights for moderation teams. This study also addresses ethical considerations, ensuring fairness, reducing bias, and maintaining freedom of expression. The research highlights the unequal influence of hate speech propagators and emphasizes the importance of personalized strategies to combat their impact. By providing a scalable and dynamic platform for ranking hate users, this study contributes to the ongoing efforts to combat online hate, supporting moderation efforts and promoting a healthier digital environment