Mapping global sentiments and geopolitical nuances in the Russia–Ukraine war through Facebook narratives

This study examined how Facebook users have expressed sentiments and political bias on the ongoing Russia–Ukraine war. The VADER model and LDA-based topic modeling were employed to analyze 130,133 Facebook posts related to the war over a two-year period. Furthermore, the XGBoost classifier model was used to annotate data on political bias toward Russia or Ukraine. The study found strong polarization, indicating posts were 65.4% negative, 12.4% positive, and 22.3% neutral. While significant proportions of positive and negative discussions (74% and 62%, respectively) showed a pro-Ukraine bias, most neutral discussions (approximately 71%) were leaning toward Russia. The prevalence of negative sentiment suggested that a significant proportion of the data expressed criticism or disappointment, whereas a relatively minor proportion of positive and neutral sentiments were either supportive or did not take either side. The political bias category within 10 countries showed significant differences in sentiments analyzed across different regions.