Assessing the Suitability of Social Media Data for Identifying Crisis Events in Smart Cities: An Exploratory Study on Flood Situations

Abstract

Social media have been used to extract different types of information. The objective of this exploratory research was to investigate the characteristics necessary to identify crisis events with social media, analyzing the aptitude of messages produced in three crisis events that hit cities in the southeastern region of Brazil. In total, 3,042 Twitter posts were analyzed based on three essential dimensions for event identification: semantic, temporal and geographic information. The results show that users actually write messages about urban floods in Brazilian cities. However, it is possible to observe differences in volume, agility and location properties posted in areas with different numbers of populations. In addition, most posts lack information. Naturally, this limits the automatic use of these messages. Before applying some automatic detection technique, managers can investigate the data that circulates in a certain region and employ strategies to improve the use of this data.