Role of Geolocation Prediction in Disaster Management

Once a disaster occurs, the common practice nowadays is that people check social media platforms, where the news usually breaks, to find out up-to-the-minute situational updates. In fact, news agencies do likewise, not only individuals. Among the important information that is needed during disaster events is geolocation information (e.g., where the disaster event has happened, where affected people are now, etc.). Such information plays an essential role in disaster management for affected people and also for response authorities such as the Intergovernmental Organizations (IGOs) and Nongovernmental Organizations (NGOs). It helps affected people to share accurate updates on their status, their needs, and the emerging incidents, which enable a rapid response. Furthermore, the geolocation information allows response authorities to manage their response activities (e.g., routing rescue teams), and reduce the impact of disasters by planning future activities (e.g., evacuation). This chapter links stakeholders’ requirements with existing computational methods for geolocation inference and introduces the computational tasks that fulfill stakeholders’ unmet needs. It also discusses the Location Mention Prediction (LMP) task due to its key role for tackling all geolocation tasks. Moreover, it discusses different categories of challenges associated with LMP subtasks, reviews the existing solutions for each and their drawbacks, and sheds light on a few future directions.