Indonesian Public Sentiment on Disasters: A Topic Modeling Study Using Twitter Data

Abstract

This study analyzes Indonesian public responses to natural disasters using Twitter data through Topic Modeling (LDA) and Sentiment Analysis. The process includes data collection, preprocessing, modeling, and sentiment classification. Tweets with the keyword “bencana” were scraped and prepared for analysis. LDA revealed 15 main topics covering disaster causes, impacts, responses, and resilience. Sentiments were categorized as positive, neutral, or negative. Positive sentiments dominated, especially regarding collaboration and response efforts. Neutral sentiments discussed resource distribution and awareness, while negative sentiments focused on dissatisfaction with disaster management and environmental issues. The study offers insights into public discourse on disasters and supports improved response strategies, awareness programs, and policy planning. This research uses NLP techniques to capture public concerns and highlights the need to strengthen community resilience in disaster situations.