PolarWeather: Visual Analysis of Multi-Platform Polarization Dynamics on Social Media
Authors
Jianing Yin (Zhejiang University), Tan Tang (Zhejiang University), Haobo Zheng (Zhejiang University), Buwei Zhou (Zhejiang University), Lu Ying (Zhejiang University), Songela Nurdawuliet (State Key Lab of CAD&CG, Zhejiang University), Yuan Tian (Zhejiang University), Tai-Quan Peng (Michigan State University), Dazhen Deng (Zhejiang University), Yingcai Wu (Zhejiang University)
Presentation
- Session
- My followers need to know about this!
- Time
- Thursday, Nov 12, 10:00 – 10:12 (US/Eastern) · session 10:00 – 11:30
- Location
- Hall America center
Keywords
Polarization dynamics analysis, social media visualization, visual metaphor, multi-platform analysis
Abstract
Polarization dynamics analysis seeks to understand how opposing stances on specific issues and the hostility between them evolve over time. These evolutions become particularly complex in today's multi-platform social media environment, where shifting topics, platform-specific polarization patterns, and cross-platform influence intertwine. Existing studies typically focus on single-platform scenarios or merely rely on macro statistics, oversimplifying fine-grained evolution patterns. To bridge the gap, through expert collaboration and a literature review, we characterize polarization dynamics at a detailed level and systematically summarize underlying factors that can inform hypotheses about possible mechanisms. To effectively visualize these detail-level dynamics, we propose a novel weather-inspired visual metaphor, in which temperature (encoded by colors) represents stances, air masses depict polarization dynamics, and fronts illustrate interaction patterns between dynamics. The metaphor is implemented by our polarization visualization algorithm, which places multi-platform polarization dynamics with a hierarchical force-directed layout and visualizes their intensity and interaction patterns with a stance-based scalar field. We further present PolarWeather, a visual analytics system that enables experts to explore multi-platform polarization dynamics and conduct factor-informed hypothesis generation. We validate the effectiveness and usability of PolarWeather through real-world case studies, expert interviews, and a user study.
For Practitioners
Practitioners in social media analysis, polarization analysis, media and communication analysis, and visual analytics design would be interested in this paper. PolarWeather demonstrates how multi-platform social media data can be analyzed from topic-level trends to stance-aligned groups, their evolution, confrontations, and potential cross-platform influence. Social media and polarization practitioners could apply this workflow to compare platform-specific polarization patterns, identify representative narratives and conflict patterns, and generate hypotheses based on various underlying factors. Visualization practitioners could adapt the paper's weather-inspired visual metaphor and coordinated overview-to-detail workflow when designing analytical interfaces for other complex, evolving, and multifaceted data.