Data Warehousing for Social Networking and Human Mobility Research and Detecting Real-world Events in Space and Time Using Feature Clustering

by Alec Pawling In this dissertation, we address two problems associated with the development of an emergency response system that utilizes a cell phone network as a sensor network to facilitate the initiation and online validation of predictive simulations. Schoenharl [77] describes the simulation component of the emergency response system; this dissertation addresses two other significant components of the system: the historical data source and the detection and alert system. The historical data source is a data warehouse designed to facilitate the development of additional simulation models for the predictive component of the system and has wider applications for scientific research on social networks and human mobility. The detection and alert system provides an automatic mechanism for initiating the simulation system without intervention by emergency response managers. This system identifies potential emergency events in both time and space, allowing the simulations to begin shortly after an event as well as focus on the area affected by the event, reducing the computational costs significantly.