Cyber Social Threats: A Data-centric AI Perspective

With the proliferation of social media platforms, cyber social threats such as fake news, hate speech, and cyberbullying have increased significantly. With the availability of abundant data for building machine learning models, artificial intelligence (AI) is making a deep impact in almost every domain, including detecting and mitigating cyber social threats on social media platforms. As the role of data in AI has significantly intensified in recent years, the concept of data-centric AI has emerged, gradually shifting the research focus from improving model design to enhancing the quality and quantity of data. In this paper, we first present the limitations of model-centric AI in the context of cyber social threats, then discuss the existing literature that uses data-centric AI techniques to detect and mitigate cyber social threats. We finally present the major challenges that social media data presents to deal with cyber social threats using data-centric AI and describe future research opportunities.