Text mining hurricane harvey tweet data: Lessons learned and policy recommendations. (15th February 2022)

• Record Type: Journal Article
• Authors: Ngamassi, Louis; Shahriari, Hesam; Ramakrishnan, Thiagarajan; Rahman, Shahedur
• Is Part Of: International journal of disaster risk reduction. Volume 70(2022)
• Publication Date: 2022-02-15
• Subjects: Disaster management -- Disaster relief recommendations -- Tweet data -- Latent dirichlet allocation (LDA); Emergency management -- Periodicals; Risk management -- Periodicals; Disaster relief -- Periodicals; Hazard mitigation -- Periodicals; 363.34
• DOI: 10.1016/j.ijdrr.2021.102753
• Languages: English
• ISSNs: 2212-4209

Abstract

Abstract: During any crisis, relief efforts depend on the timely exchange of crisis-related information between organizations and the communities. Existing literature shows that relief efforts often fall short in terms of effective communication. One of the possible reasons for this is a misalignment between the expectations of people and the efforts of disaster respondents. Thus, understanding people's needs and expectations during disasters may help reduce the gap between the key stakeholders. In this paper, we use the Latent Dirichlet Allocation (LDA) technique to mine tweet data collected during Hurricane Harvey to better understand the needs of the people in the disaster area. Through data mining, we identify five themes of concern by Twitter users during the pre-crisis period of Harvey: disaster declaration and emergency response, concern about a specific town, event or travel cancellations, threat to oil & gas (energy) industry, and climate change. Based on these themes, we provide recommendations to help disaster management agencies and policymakers be better prepared to assist disaster victims and facilitate citizens' involvement.