Mining and analysis of public sentiment during disaster events: The extreme rainstorm disaster in megacities of China in 2021

Abstract

Cities are concentrated areas of population that are vulnerable to the impact of natural disasters. Owing to the impact of climate change and extreme weather incidents in recent years, many cities worldwide have been affected by sudden disasters, especially floods, causing many casualties. Social media plays an important role in the communication and sharing of information when physical communication is limited in emergency situations. However, obtaining and using public sentiment during major disasters to provide support for emergency disaster relief is a popular research topic. In the summer of 2021, China's inland plains experienced extremely serious rainstorms. The rainfall on July 20 in the capital city of Zhengzhou, Henan Province, the most population province in China, reached 201.9 mm/h, causing extremely serious consequences. This case study examines people's sentiment about this event through datamining of Chinese Weibo social media during the extreme rainfall period. The six most concerned types of public response topics and 14 subcategory topics were determined from 2,124,162 Weibo messages. “Asking for help” and “public sentiment” dominated the main topics, reaching almost 66%, with a relatively even distribution of secondary categories, but with “appeal for assistance” taking the top spot. Topics changed cyclically with work and rest, but these areas seemed to lag behind coastal areas in their responses to the storm in the same time. The topics were centred around Zhengzhou and distributed in China's major city clusters, such as the Beijing-Tianjin-Hebei agglomerations, Yangtze River Delta, and Pearl River Delta regions. Community-level disaster relief information was also discovered, which showed that high building power outages, basement flooding, tunnel trapping, and drinking water shortages were common topics in specific inner urban regions. This detailed information will contribute to accurate location-based relief in the future. Based on this lesson, a series of measures for urban flood reduction are proposed, including disaster prevention awareness, infrastructure building, regulation mechanisms, social inclusivity, and media dissemination.