Doctoral Dissertation Research: Task Performance and Causal Claims: Leveraging Network Analysis and Large Language Models to Extract Information from Organizational Hazard Texts

Crises – particularly those rising to the level of disasters –demand an organized response. In complex systems of organizational disaster response, core tasks such as communication and coordination of activities become large-scale challenges. This study evaluates such challenges in two research thrusts.

The first thrust examines the broader discourse created by official online communications disseminated to the public during the COVID-19 pandemic. The second thrust examines the extent to which an organization's task performance impacts effective collaborations in an emergent multi-organizational disaster response network, with a specific case study of the 2005 Hurricane Katrina.

The “lessons learned" from our communications studies are used to synthesize concrete organizational communication strategies that can help practitioners and officials better engage stakeholders, thereby building the trust between the public and officials that is critical for disaster management. The results of our collaboration study inform organizational training efforts for hazard events, which may decrease the friction of collaboration and coordination efforts in response to large-scale disasters and thus contribute to the protection of lives and property.

This study aims to provide insights at multiple levels of the organizational response process, from shared task performance among organizations to the broader structure of public-facing discourse around health hazards and the emergence of multi-organizational response networks. This study leverages the strengths of machine learning-based natural language processing methodologies and network analysis techniques to extract from, and analyze massive corpuses of hazards communications.

Thus, this study contributes to methodologies of large-scale information extraction, decreasing costs previously associated with obtaining high-quality data from records while increasing the potential value of underutilized historical case studies.

This award reflects NSF's statutory mission and has been deemed worthy of support through evaluation using the Foundation's intellectual merit and broader impacts review criteria.