<![CDATA[Sumatra Flash Flood triggered extensive public discussion on social media regarding the causes of the disaster, particularly between natural factors and human factors. Understanding public perception is important for evaluating disaster literacy and mitigation communication strategies. This study analyzes public perceptions of the disaster’s causes using sentiment analysis based on the Long Short-Term Memory (LSTM) algorithm applied to YouTube comments. A total of 15,259 comments were collected using the YouTube Data API and processed through data cleaning and lexicon-based automatic labeling. After data selection, 11,043 comments were used for model training and testing. The experimental results show that the LSTM model achieved an accuracy of 96.29% and a ROC-AUC value of 0.99. Sentiment distribution analysis indicates that 57% of comments emphasize human factors, while 43% highlight natural factors or religious interpretations. These findings suggest an increasing public awareness of the human role in disaster risk and demonstrate the effectiveness of LSTM for Indonesian-language disaster sentiment analysis.]]>
Copyrights © 2026
