Yields Valuable Information for Emergency Responders
Many would rightly argue that the ability to access real-time information as a disaster unfolds—and throughout the event’s duration—is critical to the efforts of emergency command and first responders to mitigate damage and save lives.
CCICADA—through its partners at Rutgers University, Carnegie Mellon University, University of Southern California-Information Sciences Institute, Rensselaer Polytechnic Institute and University of Illinois at Urbana Champaign—is researching data-science applications to initiate real-time situational awareness and emergency response through the use of social media. This research focus was motivated by discussions with US Coast Guard leadership and interest from other homeland security entities such as the First Responders Group at the US Department of Homeland Security, Science and Technology Directorate, and at FEMA.
Social media can provide a valuable source of information to those who need to make good decisions about how to respond to an emerging disaster. This information must be combined with sensor data, computer output, radar, and many other sources, each contributing in its own way to the developing understanding of what is happening and of the options for response. However, social media has great promise as a way to complement the more traditional tools for gaining situational awareness.
CCICADA’s social-media, situational-awareness research is analyzing the links between social media message characteristics and human behavior in response to actionable information during an emerging event. For example, an explosion onboard a passenger ferry is witnessed offshore. In the early phases of the event there is much uncertainty and confusion about its cause and the extent of damage. This uncertainty delays the type and level of the response and hampers the delivery of needed recovery services.
However, there is likely a source of key information that should not be overlooked: social media, such as Twitter or Facebook. Ferry passengers and crew may communicate through social media, giving emergency-response teams a time-referenced inside view of what has happened, where it is located, whether it is ongoing, what injuries have occurred, and the like. External witnesses, such as those on shore or on other vessels, may provide further social-media information about the event.
CCICADA has conducted several research studies—for example, on the use of Twitter during the 2010 Haiti earthquake, 2011 Japanese tsunami and Superstorm Sandy in 2012—which offer insights into related human behavior during the events. The studies showed how extreme events with different characteristics can prompt different human behavior if the use of social media is involved. The research also showed that social media discussions around a disaster follow a largely predictable series of stages. Understanding these stages can provide insight into the status of an event as it is unfolding.
CCICADA is also developing methods to collect, refine, and display information from social media to support emergency response planning and management activities. These methods identify social-media users who are present at the event vs. people merely messaging from a distance.
These methods also identify the most informative and active message authors, specifically with an eye to establishing connections with them during a crisis. This work connects with earlier CCICADA work on “trustworthiness estimation.” Information from social media is useful to gain situational awareness only to the extent that one can trust that information.
In another component of this work, CCICADA is enhancing prior research into the 2010 Haiti earthquake, creating a framework that can group social media messages by location, determine top requests by location, label aid requests (medical, food/water, transportation), and help allocate aid based on traditional methods and pre-existing facility locations that integrate the social media geo-locations and requests.
CCICADA is now developing methods to identify the location of the social media where no such information is explicitly mentioned in the communication. This will provide additional emergency data for responders, enabling them to separate the data points addressing the emergency from a distance.