Advancing social insights through NLP: social media reactions, mental health and beyond

Understanding the intricate relationship between digital communication, public policy, and societal well-being is increasingly important in our interconnected world. Traditionally, analyses in these areas have relied heavily on first-hand data collection methods, such as surveys, which can be time-consuming and limited in scope. This thesis aims to transcend these limitations by leveraging the vast and diverse data available on social media platforms. By applying advanced natural language processing techniques to social media content, the research provides novel insights into how digital discourse shapes public opinion, affects policy efficacy, and influences individual well-being.

The thesis harnesses the richness of social media data to explore the politicization of policy discussions, the mental health implications of major societal events like the COVID-19 pandemic, and the growing issue of loneliness in the digital age. We demonstrate how politicized narratives in digital platforms can skew public perception and policy effectiveness. Furthermore, we delve into the psychological impacts of the pandemic, revealing significant variations in stress, anxiety, and loneliness across different demographic groups, with a notable focus on younger women. Finally, we introduce an innovative approach for traditional machine learning models to detect and analyze expressions of loneliness on social media, offering potential pathways for timely and effective mental health interventions. The outcomes of our findings help better understand the complex dynamics between digital communication and key societal challenges, offering invaluable insights for policymakers, mental health professionals, and the general public.