Understanding RSM: Relief Social Media
September 15, 2009
Microsoft PowerPoint - Need a reader? Get one here
This presentation describes a new ONR-sponsored two-year research project
'Understanding RSM: Relief Social Media' that we are beginning in
conjunction with colleagues at the Lockheed Martin Advanced Technology
Laboratory. The RSM project is aimed at helping to detect and monitor
information about crises and associated relief efforts from online
sources including both social media and main stream media.
There has been a dramatic rise in the use of online social media systems like Twitter, Facebook, and Flickr are increasingly being used to describe and document events related to natural disasters, political turmoil, terrorist attacks, pandemics and other crises as they unfold.
Often these sources provide the only reliable stream of timely information about what is actually happening on the ground. Mobile phones are being used to capture and upload pictures, audio and videos which are often annotated with metadata accurately capturing the time and location as well as semantic tags. At the same time, citizen and professional responders in relief scenarios are using the same technologies to report data and coordinate their activities. In effect, these systems enable the population to act as a human sensor network.
The project has three tasks. The first is to build a system to harvest empirical evidence about crises, resulting in a unique cross-system data corpus, whose empirical characterization can be used to guide further research and engineering. The second task will focus on creating a computational model to codify our theoretical understanding of the dynamics of these systems. A software simulator of the abstract model will be used to generate virtual network instances to experiment against. Finally, we will develop a analysis toolkit that will allow us to support forecasting and planning of decision makers. In particular, we can generate appropriate confidence levels and model both information source influence and bias in content from open sources.