A systematic review of social media-based sentiment analysis in disaster risk management

This study seeks to enhance understanding of how social media-based sentiment analysis can contribute to disaster risk management. As a natural language processing tool, sentiment analysis is increasingly used to extract public emotions, concerns, and responses during disasters. This systematic review analyzes 139 studies and employs the PRISMA framework for literature search and selection. It uses both inductive and deductive techniques for content analysis. The review offers an overview of the field, identifying key themes and prominent tools adopted by researchers. It reveals that various techniques, including machine learning algorithms, lexicon-based methods, and hybrid models, are employed to assess public sentiments across platforms such as X, Weibo, and Facebook. Furthermore, the review highlights the merits of social media and sentiment analysis and the significant limitations of these platforms and sentiment analysis tools. The review reveals that despite limitations, social media is a meaningful resource for gauging societal sentiments for all phases of disasters. It finds that sentiment analysis uncovers emotions essential for devising effective strategies in disaster risk management. A major gap in the existing literature is the reliance on English and Chinese sources, leading to research skewed toward data-rich countries. Another gap is the predominant use of platforms such as X and Weibo. The findings suggest the need to consider broader language inclusivity, geographic diversity, and the use of multiple data sources to enhance the effectiveness of sentiment analysis for disaster risk management. Insights from this study can be used to create more resilient communities in the face of climate change and other stressors. • Literature on social media-based sentiment analysis for DRR is reviewed. • Key themes and prominent tools adopted by researchers are identified. • Social media could be a meaningful resource for gauging sentiments for enhanced DRR. • Research is skewed toward data-rich countries and focused on English and Chinese sources. • Issues such as variations in writing style and privacy concern need to be addressed.