Study Analyzes Brazil's YouTube Vaccine Discourse Through 1.4 Million Comments

A study focused on Brazil's vaccine discussions on YouTube analyzed approximately 1.4 million comments through a semi-supervised stance detection approach. This research fills a gap in understanding the long-term trends in non-English settings, specifically Brazil's extensive immunization framework. By merging stance classification with engagement metrics, channel taxonomy, and temporal patterns, the study illustrates the polarization between pro- and anti-vaccine sentiments. It emphasizes how misinformation online, political divisions, and waning trust in institutions can hinder vaccination efforts, particularly evident during the COVID-19 crisis. Unlike previous computational analyses that mainly dealt with English data or specific vaccines over short periods, this methodology employs self-labeling and self-training to assess comments from various media sources, offering valuable insights into Brazil's media landscape and its influence on public health discussions.