Graduate Seminar: Towards Analyzing Online Communities of Problematic Information: A Computational Approach

Abstract

Problematic information–information that is inaccurate, misleading, inappropriately attributed, or altogether fabricated–prevails in digital spaces. Engaging with problematic information online can lead users toward a path of social distrust, paranoia, and radicalization. Understanding how users traverse the web of problematic information, and how they find ways out of it is crucial to reducing the spread of disinformation and hate online.

Despite the obvious disruptive consequences of online communities of problematic information, we don't have a large-scale, data-driven understanding of deeper social processes that are underway. In this talk, I discuss my research contributing to a multidisciplinary understanding of social processes of engagement, mobilization, and disengagement from problematic information online. My computational research is grounded in theories of social psychology and complemented by qualitative methods, spreading across multiple social media platforms such as Reddit, Facebook, Twitter, and 4chan. In this talk, I focus on three main themes:

• What makes people engage in conspiracy theory discussions?
• What could motivate people to leave online conspiracy theory discussions?
• What are the mechanisms of information mobilization in online hate and influence movements?

Building on my past research, I will also briefly outline my future agenda for designing fair, ethical, and civil interventions for online public discussions containing problematic content.

Speaker Biography

Shruti Phadke is a Trust and Safety researcher at TikTok USDS and a visiting researcher at the University of Texas Austin. She received a Ph.D. in Information Science from the University of Washington Information School in June 2023. Her research focuses on computationally understanding participation, social knowledge construction, and resource mobilization in online communities of problematic information. Her research benefits from multiple methodological approaches ranging from statistics, causal machine learning, natural language processing, and qualitative methods. Her research has received two Best Paper Honorable Mention awards at CSCW and the Best Paper Award at ICWSM 2022.