Multidimensional report analysis in urban incident management

Abstract

In urban incidents and crises, accurate and timely information can be crucial to manage critical situations. The exponential growth of crowd sourced data has given means to access vast amounts of information on a real-time basis. However, this has complicated the task of analyzing ongoing events as the effort needed to filter relevant from irrelevant information has exponentially grown. This paper proposes a multidimensional analysis method of processing high influx of crowd sourced incident reports and creating processable pieces of information by filtering what's irrelevant and clustering what belongs together in a highly efficient way. Spatial, temporal, and semantic dimensions of an incident report constitute a basis which is taken advantage of in this work to ease the tasks which are undertaken manually in operation centers and alike.