Friday, August 12, 2011
Novel Social Networking in Disaster Management
The Disaster Management Initiative at Carnegie Mellon Silicon Valley is harnessing the social media revolution to develop tools for disaster response. Recent success stories highlight the potential of technological aids in disaster response, from the use of crowd-sourcing to aid translation after the Haiti earthquake, to volunteer mappers aiding emergency managers in the San Bruno Fire, to helping categorize and geo-locate messages from the ground during flooding in Pakistan so decisions-makers could make sense of the data in a map format.
Mapping messages and photos is essential for creating a working Common Operating Picture (COP) from social media. As people at the site of a disaster send text messages and post to sites such as Twitter and Facebook, the volunteers in groups such as CrisisCommons are geo-locating messages, which aid decision-makers in assessing a situation and triage resources without being onsite. There is great potential for social media to provide real-time status of infrastructure that could be used, for example, to alert first responders in advance if the route to a location is blocked by a collapsed bridge.
Issues preventing that from happening include being able to rapidly find relevant “microblogs” (such as Twitter feeds) on situational awareness. Challenges center on terminology, around the volume of the Twitter stream, different social media platforms, language barriers, concerns over accuracy of social media reports, and more. For emergency managers, accuracy of information is a key issue. Misinformation in reports can cause a myriad of problems and delays in effective crisis management. DMI researchers are looking into ways to help identify accurate reports, and correct inaccurate ones, using a variety of methods, including corroboration, context, meta data, and Bayesian decision networks.
Jeannie Stamberger, Associate Director at CMUSV’s Disaster Management Initiative, is applying expertise in modeling and statistical analysis of patchy data to address issues and explore opportunities in using social media in disaster response. Dr. Stamberger participated in an earthquake drill at Stanford University, monitoring Twitter feed while the entire campus was being evacuated. The experiment provided insights into breadth of information and terms people used during the drill, allowed interactive question and response between managers and participants, and provided novel insight relevant to the emergency managers.
The DMI is assisting local municipalities and emergency response organizations seeking effective ways to use social media to deliver information and alerts to the public. CMUSV faculty and students are conducting related research, working on modeling of general characteristics of tweeting behavior. Research is collating best practices, which has revealed the fact that many reporting agencies max out the 140 character limit when posting messages on Twitter. That, in turn, leaves no room to spread the message through retweeting because the retweeter has to cut the message, which makes it look like the retweeter has tampered with the message.
In one research study PhD students applied statistical modeling to study why people retweet on Twitter. They analyzed user’s retweeting behavior by studying factors that may affect the decision, including context influences, network influences and time decaying factors. “The findings from this research, which models the retweeting process, shed light on how government agencies can strategize their tweeting behavior in order to best spread information to the general public in emergencies and disasters,” said supervising Professor Joy Zhang.
One benefit of using social media is its built-in resiliency. People don’t have to learn a new software program in the midst of a crisis. Instead, they can aid disaster response efforts with a tool they use every day. Events such as The Great California ShakeOut demonstrate the public’s ability to use the tools available to send messages and pictures. However, the public needs guidance in using those tools. DMI leaders are working with policy makers, both government and non-profit, to guide the public in responding to disasters, based on experience with social media and technology.
With cross-sector participation from practitioners, researchers, technologists, organizations, developers and volunteers, the DMI brings shared experience and resources to develop and integrate technology for disaster management. The consortium of DMI leaders and researchers provides novel insights into social media as a tool for 21st century disaster response aid.