CRITICAL ANALYSIS OF SOCIAL MEDIA NARRATIVES ON DISASTER RESPONSE: THE 2022 CIANJUR EARTHQUAKE CASE IN INDONESIA

This study analyzes social media narratives surrounding the 2022 Cianjur earthquake in Indonesia to understand public sentiment toward disaster response efforts. Using a comprehensive methodological approach, data was collected from Twitter, Facebook, and Instagram and processed through sentiment analysis using Logistic Regression and Naive Bayes models. The analysis categorized public responses into six sentiment classes: Support, Criticism, Rejection, Approval, Rebuttal, and Spam, while also mapping emotional distribution across Joy, Surprise, Fear, Sadness, Disgust, and Anger. Results reveal that Support was the predominant sentiment category, with approximately 4,700 instances identified by Logistic Regression and 4,000 by Naive Bayes, indicating overall public satisfaction with response efforts. Emotional analysis showed Joy (2,047 instances) and Surprise (1,691 instances) as dominant emotions, reflecting relief associated with survival and rescue operations. Both models demonstrated high accuracy, with AUC scores of 0.868 and 0.866 respectively, though they faced challenges in classifying minority sentiment categories. The findings suggest that real-time social media sentiment analysis offers valuable insights for disaster management, enabling authorities to quickly adapt communication strategies to address public concerns and misinformation. This research contributes to disaster preparedness frameworks by providing a methodological foundation for leveraging social media as a tool for gauging public sentiment during crisis events, ultimately enhancing community resilience and response effectiveness.