Rapid assessment of building collapse based on sequential dynamic fusion of multi-source disaster information from news media

Abstract

News media provides time and data guarantee for rapid post-disaster assessment. However, disaster information from news media often suffers from semantic ambiguity and conflicting content, and disaster information changes over time. These characteristics of disaster information seriously affect its application in disaster assessment. How to obtain timely reasonable quantitative disaster assessment results based on this kind of disaster information is the key to making full use of media information, but existing studies rarely have addressed the issue. This paper proposes a novel method to assess the percentage of buildings collapse by leveraging timely disaster information with geographical positions. The method includes four parts. First, based on preprocessing of disaster information, the cloud model is used to quantitatively express fuzzy disaster information and construct basic probability assignments (BPAs) for different ranges of buildings collapse. Second, the conflict and ambiguity of each item of disaster information is analyzed, and its credibility is measured. Based on this, the BPAs are modified to reduce conflicting disaster information. Third, multiple items of disaster information are fused by using Dempster–Shafer theory to obtain buildings damage assessment results. Finally, the assessment results of the previous stage are dynamically updated with new information. The method is applied to Ya'an earthquake on April 20, 2013 and Yi'bin earthquake on June 17, 2019. The assessment results are analyzed and compared with those obtained by other methods, and it is found that the method can obtain reliable results in a short time. This study can provide a reference for timely emergency decision-making.