Understanding the situation distribution is a fundamental but important step in the emergency response to disaster. There are various emergency related spatial data available on Internet; however, it is still a big challenge in clustering the dynamic big spatial data. In this study, we provide a dynamic spatial clustering (DSC) to efficiently load and cluster the spatial big data based on a hierarchical-partition model (HPM). We have modeled the DSC to understand the distribution of emergency (e.g. Kumamoto earthquake in May 2016) from spatial data in tweets. The major contributions in the HPM-based DSC include loading dynamic big spatial data with optimal utilization of external memory, and rapid clustering to detect the dense regions of targeted emergency.
展开▼
