Publications

This study examines how mainstream Taiwanese newspapers portrayed LGBTQ+ populations from 2010 to 2021, a period marked by rising LGBTQ+ visibility and major legal change. Drawing on institutional mediation and legitimacy perspectives, we test three hypotheses: first, that coverage becomes more negative as LGBTQ+ issues gain visibility over time; second, that gay men receive more negative coverage than lesbian women; and third, that major legal and social milestones shift coverage toward more neutral portrayals and a broader range of topics. To evaluate these expectations, we analyze 18,558 articles from China Times, Liberty Times, and United Daily News using web-scraped data, GPT-4-assisted classification, and time-series regression models. The findings reveal an overall increase in positive news coverage over the decade, but disparities in the frequency and sentiment of coverage between gay and lesbian subjects persist. The study concludes that while there has been progress toward inclusiveness, significant events and milestones have only partially influenced the portrayal of the LGBTQ+ community. The results highlight the ongoing need for efforts toward equitable media representation.

Contested narratives on social media often spread in bursts: most remain localized, yet a few tip into rapid diffusion and reshape polarized network structure. We develop an agent-based model of competing narrative diffusion on an adaptive network to explain when such tipping dynamics occur. In the model, sharing produces social feedback that reinforces conviction, adoption is modulated by time-varying collective attention, and ties coevolve as agents rewire based on disagreement. We also include committed seeders (“zealots”) to represent persistent promotion. Across simulation experiments, we identify two recurring diffusion pathways. Under strong social pressure and highly adaptive networks, diffusion exhibits rapid, localized bursts that consolidate ideologically homogeneous clusters, accelerating tipping while restricting cross-group reach. Under more moderate network adaptation, diffusion proceeds more gradually yet reaches a broader portion of the network through bridging ties, producing wider reach with less extreme segregation. Attention decay creates a finite window for large-scale diffusion; nonetheless, once established within polarized clusters, diffusion can persist at a nontrivial level even as attention wanes when zealots continue seeding. Finally, asymmetric zealot deployment across groups generates divergent outcomes, with one group consolidating into a more homogeneous cluster while the other becomes larger but more internally heterogeneous. The model offers a general mechanism for bursty narrative diffusion in coevolving networks, with implications for disinformation campaigns as an application context where strategic seeding and identity-consistent reinforcement may be especially salient.

The pervasive presence and influence of political bots have become the subject of extensive research in recent years. Studies have revealed that a significant percentage of active accounts are bots, contributing to the polarization of public sentiment online. This study employs an agent-based model in conducting computer simulations of complex social networks, to elucidate how bots, representing diverse ideological perspectives, exacerbate societal divisions. To investigate the dynamics of opinion diffusion and shed light on the phenomenon of polarization caused by the activities of political bots, we introduced bots into a bounded-confidence opinion dynamic model for different social networks, whereby the effects of bots on other agents were studied to provide a comprehensive understanding of their influence on opinion dynamics. The simulations showed that the symmetrical deployment of bots on both sides of the opinion spectrum intensifies polarization. These effects were observed within specific tolerance and homophily ranges, with low and high user tolerances slowing down polarization. Moreover, the average path length of the network and the centrality of the bots had a significant impact on the result. Finally, polarization tends to be lower when humans exhibit reduced confidence in bots. This research not only offers valuable insights into the implications of bot activities on the polarization of public opinion and current state of digital society but also provides suggestions to mitigate bot-driven polarization.