Using crowd sourced content to help manage emergency events

Abstract

The Emergency Situation Awareness (ESA) tool provides crowd-sourced information in near-real-time from Twitter about all-hazard events for emergency managers. ESA currently collects tweets from Australia and New Zealand and processes them to identify unexpected incidents, to monitor ongoing emergency events and provides access to an archive to explore past events. It is operated using a map based interactive web site and has processed over 2 billion tweets since September 2011. ESA has been developed by the Commonwealth Scientific and Industrial Research Organisation (CSIRO) and has been trialed by numerous emergency services organisations throughout Australia. Tweets are processed as a data stream using text mining techniques and natural language processing tools to identify content relevant to emergency managers. ESA is deployed as a distributed information architecture consisting of a combination of commodity open source technologies, such as an Apache Solr index, a relational database, messaging infrastructure, web servers and supporting software toolkits, as well as purpose built components for message burst detection, event identification and notification, message classification and clustering, geo-coding and searching. In this chapter, an overview of ESA is presented showing how tweets are gathered and processed. Three case studies are outlined explaining how ESA is used to detect earthquakes, monitor bushfire events and as a general all-hazard monitoring tool in a crisis coordination centre. We also note some of the issues we have encountered from using our tool and present an overview of our research road map noting the planned extensions and new features.