Authors
Mel Mason, Reynold Greenlaw, Andrew Muddiman, Matthias Moi, Thomas Ludwig, Christian Reuter, Massimo Cristaldi, Federico Sangiorgio
Abstract
This document gives a brief overview of data mining as a broad discipline, the nature of social media data and the data currently available to EMS. In order to build a working EmerGent system, data mining will have to take the results of the data gathering process and support the Information Quality and Information Routing processes. To do this three key challenges are identified and discussed:
1. reducing high-volume, low-quality social media reports into low-volume, high-quality data appropriate for further mining,
2. identifying networks, especially social networks, within the data to understand the context of an emergency,
3. to identify, categorise and match requests for, and offers of, help and supplies during and after an emergency. Methods are proposed for each of these three areas and they are then set within a larger Knowledge Discovery Process.
Purpose of the document
The main purpose of this document is to give grounds for the initial choice of data mining in EmerGent. The document gives a survey of data mining techniques and their application to social media and the information needs of emergency services. As part of the development of information mining methods (T4.2) this deliverable contributes directly to objective O3 (to identify requirements, to implement, and to evaluate methods and tools for novel emergency management in social media generation). These methods have the potential to transform high-volume, low-information social media emergency data into low-volume information-rich data for EMS.