Extracting and classifying typhoon disaster information based on volunteered geographic information from Chinese Sina microblog

Abstract

SummaryThe notion and application of volunteered geographic information occur and rise rapidly in recent years with the thriving of social media in China such as Sina Microblog, which is one of the most active social network sites. Many researches on natural disasters like flood, earthquake, and forest fires leverage social media like Twitter, Flickr, or YouTube, but few studies focus on typhoon disaster based on Sina Microblog even though typhoon disaster batters southeast coastline of China every year. This study proposed a method to extract and classify typhoon disaster information from Sina Microblog. KNN (k‐nearest neighbors) algorithm is implemented for microblog classification in order to extract useful information about the real hazards, and an experiment is conducted to tune the parameters in KNN by comparison of outcomes of social media data analysis and the real typhoon situation. The result shows that more than 70% microblogs are classified correctly. After the classification, we carried out spatial temporal analysis to map the disaster situation. It shows that the spatial distribution of microblog message mean centers about typhoon has regular variation along with the typhoon path. It can be confidently concluded that Sina Microblog has some potential prospects for estimating the typhoon disaster situation.