Context-aware spatio-temporal event

Abstract

Massive spatio-temporal event data sets, which include occurrences like disease epidemics, armed conflicts, and crimes, are now accessible. For many applications, from disease prevention to global politics, foretelling such occurrences and identifying the underlying triggering patterns are essential tasks. The spatio-temporal linkages between events are captured by conventional event prediction models based on Hawkes processes, but they are unable to take into account complicated and diverse external variables such as population distribution, weather, and topography. This article suggests an event prediction technique that efficiently makes use of the valuable external information found in collections of unstructured data (e.g., map images, satellite images and weather map).