Measuring Belief Dynamics on Twitter
Keywords:Subjectivity in textual data; sentiment analysis; polarity/opinion identification and extraction, linguistic analyses of social media behavior, Qualitative and quantitative studies of social media, Organizational and group behavior mediated by social media; interpersonal communication mediated by social media, Text categorization; topic recognition; demographic/gender/age identification
AbstractThere is growing concern about misinformation and the role online media plays in social polarization. Analyzing belief dynamics is one way to enhance our understanding of these problems. Existing analytical tools, such as sur-vey research or stance detection, lack the power to corre-late contextual factors with population-level changes in belief dynamics. In this exploratory study, I present the Belief Landscape Framework, which uses data about people’s professed beliefs in an online setting to measure belief dynamics with more temporal granularity than previous methods. I apply the approach to conversations about climate change on Twitter and provide initial validation by comparing the method’s output to a set of hypotheses drawn from the literature on dynamic systems. My analysis indicates that the method is relatively robust to different parameter settings, and results suggest that 1) there are many stable configurations of belief on the polarizing issue of climate change and 2) that people move in predictable ways around these points. The method paves the way for more powerful tools that can be used to understand how the modern digital media eco-system impacts collective belief dynamics and what role misinformation plays in that process.
How to Cite
Introne, J. (2023). Measuring Belief Dynamics on Twitter. Proceedings of the International AAAI Conference on Web and Social Media, 17(1), 387-398. https://doi.org/10.1609/icwsm.v17i1.22154