Unveiling the spatiotemporal propagation patterns of sentiments regarding the Israeli–Palestinian military conflict

Military conflicts occur in the physical world, while conflicting opinions simultaneously emerge in cyberspace. The rise of social media has facilitated the dissemination and exchange of these opinions. This paper examines the spatiotemporal propagation patterns of sentiments expressed by X.com (formerly Twitter) users regarding “Israel and Palestine” since October 7, 2023. Approximately 151,400 tweets were collected from 146 countries and 293 cities. A BERT model was fine-tuned to classify these tweets into six basic sentiment categories: anger, fear, joy, love, sadness, and surprise. In the spatiotemporal propagation analysis, community detection in social networks was employed to explore the spatial propagation patterns of sentiments. Time series and indicator data from different geographic regions were clustered and classified to identify the main factors influencing sentiment propagation. The results indicate that (1) the sentiment propagation network is divided into multiple communities (e.g., the US-group, the France-group, and the UK-group) with clear geographic clustering characteristics, suggesting that X.com users are inclined to engage in sentiment exchanges regarding “Israel and Palestine” with others from similar regions; (2) the six basic sentiments expressed by X.com users are more frequently propagated within individual countries, while international sentiment propagation is concentrated in regions such as Northern America and Western Europe; and (3) sentiments expressed by X.com users in geographic regions with similar factors—such as economy, Internet usage, and religion—exhibit aligned patterns in time-series trends (e.g., Northern America, Europe, Australia, and New Zealand). These findings offer a global perspective for understanding the Israeli‒Palestinian military conflict.