Visual Analytics of Microblog Data for Public Behavior Analysis in Disaster Events

Abstract

In disaster management, analysis of public behavior plays an important role for evacuation planning. Unfortunately, finding meaningful information for analysis is challenging and collecting relevant data can be very costly. However, the growing dataset of Location-based Social Networks services with its time-stamped, geo-located data offers a new opportunity. Such spatiotemporal data has substantial potential to increase the situational awareness of local events and provide for both planning and investigation. In this paper, we present a visual analytics tool that provides users with interactive social media data analysis and investigation in order to help evacuation planning, analysis, and response. We demonstrate how to improve investigation by analyzing the extracted public behavior responses before and after the evacuation order during the natural disaster event, such as Hurricane Sandy.