CRISP Type 2: Collaborative Research: Simulation-Based Hypothesis Testing of Socio-Technical Community Resilience Using Distributed Optimization and Natural Language Processing

University Of Washington, Seattle WA

Investigators

Abstract

Critical infrastructures, such as electric, manufacturing, and financial systems, are key to the functioning of society and the health of communities. The new knowledge from this project will improve the design and management of critical infrastructure to build resilience in the face of minor disruptions and large disasters. The project focuses on the social and technical links between different types of critical infrastructure. The research provides insights into the influence of social forces on critical infrastructure and the roles of critical infrastructure in promoting a community's identity and well-being. The knowledge and tools generated from this research inform strategies to improve the functioning and operation of critical infrastructure in order to achieve socially defined goals. Houston and Seattle area experts and decision makers contribute to and evaluate project outcomes to ensure that resulting tools are relevant to stakeholders concerned with increasing community resilience capacity. Research methods from civil engineering, computer science, and social science combine to achieve the goals of the project. Three primary goals of the research are to 1) systematically rethink critical infrastructure as a web of social and technical systems, 2) build computer simulation models to explore critical infrastructure performance after major and minor disruptions, and 3) test hypotheses to appreciate how critical infrastructure can improve resilience and support the diverse needs of communities. The research team integrates qualitative and quantitative data to construct the project's simulation models. The scholars compile and analyze qualitative data about past critical infrastructure disruptions and disasters from many text sources, such as social media, news stories, government documents, and industry reports. They use new natural language processing (NLP) methods analyze the text data in order to identify key variables describing critical infrastructure and community resilience, as well as the relationships among them. They collect quantitative and geographic data describing related variables from existing sources, such as government and academic databases. They elicit quantitative data also from experts using customized survey techniques during facilitated workshops. They then use the results of the data analysis to specify computer models that simulate the many events, resource exchanges, and decisions that occur across multiple geographic scales after critical infrastructure disruptions and disasters. Finally, the team devises techniques to integrate and optimize the constructed computer models. This permits efficient testing of hypotheses about the relationships between critical infrastructure performance and community resilience.