Opinion Dynamics Models for Sentiment Evolution in Weibo Blogs

TL;DR

This paper investigates how follower sentiment toward Huawei-related content on Weibo evolves under influencer-follower interactions. By aggregating follower feedback into a macro-agent sentiment $\sigma_b(t)$ and evaluating multiple opinion-dynamics models, including delayed and two-layer (private vs expressed) variants, the study shows that averaging-based dynamics capture the observed trajectories, with delayed, two-layer models (notably the reduced EPO model with lag $\tau=2$) providing the best predictive performance and revealing sparse inter-blog influence structures. The findings offer interpretable, data-driven tools for forecasting collective mood shifts and identifying influential blogs, with practical implications for marketing, crisis communication, and brand management on social media. The methodology combines topic-focused sentiment analysis using a fine-tuned BERT model, careful data aggregation, and dynamical-system identification to learn influence matrices from time-series data. Overall, the work demonstrates that parsimonious, averaging-based opinion dynamics can robustly explain and predict sentiment evolution in large-scale social media communities, while highlighting when private vs expressed opinions matter for accurate modeling.

Abstract

Online social media platforms enable influencers to distribute content and quickly capture audience reactions, significantly shaping their promotional strategies and advertising agreements. Understanding how sentiment dynamics and emotional contagion unfold among followers is vital for influencers and marketers, as these processes shape engagement, brand perception, and purchasing behavior. While sentiment analysis tools effectively track sentiment fluctuations, dynamical models explaining their evolution remain limited, often neglecting network structures and interactions both among blogs and between their topic-focused follower groups. In this study, we tracked influential tech-focused Weibo bloggers over six months, quantifying follower sentiment from text-mined feedback. By treating each blogger's audience as a single "macro-agent", we find that sentiment trajectories follow the principle of iterative averaging -- a foundational mechanism in many dynamical models of opinion formation, a theoretical framework at the intersection of social network analysis and dynamical systems theory. The sentiment evolution aligns closely with opinion-dynamics models, particularly modified versions of the classical French-DeGroot model that incorporate delayed perception and distinguish between expressed and private opinions. The inferred influence structures reveal interdependencies among blogs that may arise from homophily, whereby emotionally similar users subscribe to the same blogs and collectively shape the shared sentiment expressed within these communities.

Figures (7)

• Figure 1: Sentiment analysis algorithms: ROC curves.
• Figure 2: Observed collective sentiments in seven blogs.
• Figure 3: Functions $\mu_b(t,\tau)$ under two scenarios: no delay in averaging ($\tau = 0$) and with a delay in averaging for the first three lags ($\tau = 1,2,3$). The dashed lines indicate a 10% exceedance or shortfall of the admissible values.
• Figure 4: Reduced EPO Model: Observed and two-period ahead predicted sentiments, $x^e_b(t)=\sigma_b(t)$ and $\widehat{x}^e_b(t)$.
• Figure 5: Fitted matrices $\widehat{W}$ and $\widehat{A}$ as heatmaps and their associated graphs.