Hurricane Katrina disrupted many social networks. Over 1.5 million people evacuated New Orleans. Companies, schools, and families were geographically separated after the storm, and they needed to rediscover the locations of their members. A "burst" of new data sources, generated on the web by disaster relief organizations, news outlets, employers, and hurricane survivors, aided those affected by the storm in searching for members of their social groups. This project investigates methods to automate tracking the geographical movements of social networks using the variety of information sources available on the web. As more time passes from Hurricane Katrina, these sources become increasingly inaccessible; therefore, the first step is to quickly harvest these Katrina-related blogs and forums, survivor databases and websites. Next, this project explores new link analysis techniques that utilize these new heterogeneous data sources to track the geographical movement of the network members, given information about the social networks prior to the disruption. These new techniques combine previous work in social network analysis, based on the study of the patterns of connections between members or subgraphs in a social networks, and link analysis methods, which determine whether a connection or link exist between members. The project will result in a range of broader impacts. The project will aid in harvesting and archiving Hurricane Katrina-related highly perishable data that will provide a historical record of the disaster. The resulting information resource will allow researchers to access an extensive dataset to develop new research in social networks and disaster recovery. The new link analysis developed in this project will assist in automating the social network reconnection task for Hurricane Katrina victims and make us better prepared to handle future disasters. The project Web site (www.eecs.tulane.edu/tejada/katrina) provides access to more information about the research and the results.
