Author ORCID Identifier:
Date of Graduation
12-2025
Document Type
Dissertation
Degree Name
Doctor of Philosophy in Engineering (PhD)
Degree Level
Graduate
Department
Industrial Engineering
Advisor/Mentor
Milburn, Ashlea
Committee Member
Haitao Liao, Haitao
Second Committee Member
Dagtas, Serhan
Third Committee Member
Zhang, Shengfan
Keywords
Depth–Speed Function; Disaster Response Logistics; Flooding; Large Language Models; Social Media Analytics; Time-Dependent Vehicle Routing
Abstract
Natural disasters pose significant logistical challenges, complicating emergency response due to unpredictable and often severe infrastructure damage. The effectiveness of disaster response is highly dependent on accurate and timely situational awareness, particularly regarding the status and usability of transportation networks such as roads and bridges. Traditional information sources often lack the immediacy needed during critical moments, prompting the use of social media for real-time insights, such as to identify damaged infrastructure or assess critical resource needs. However, social media data often suffer from imprecise geolocation and potential misinformation, limiting their direct applicability. This dissertation introduces a comprehensive framework that integrates images from social media with video and other data from reliable sources such as traffic cameras and satellite images to improve disaster response logistics. It utilizes advanced multimodal geolocation techniques, combining computer vision and natural language processing, demonstrated in a Hurricane Harvey case study. In addition, the research goes beyond simplistic binary road condition assessments, developing sophisticated data-driven models based on real-world vehicle and flood interaction data. These models accurately reflect the levels of road passability and vehicle speeds associated with varying road inundation levels. Finally, a novel variant of the Time-Dependent Shortest Path and Vehicle Routing Problem (TDSPVRP) incorporates dynamic road conditions, vehicle dispatch timing, and heterogeneous fleet considerations while optimizing relief distribution by minimizing maximum arrival times at destinations. The dissertation advances disaster response logistics through structured, actionable strategies leveraging advanced analytics, artificial intelligence, and optimization techniques.
Citation
Odebode, A. (2025). Enhancing Disaster Response Logistics via Social Data: An Analytics Approach to Improving Situational Awareness of Infrastructure Status. Graduate Theses and Dissertations Retrieved from https://scholarworks.uark.edu/etd/5985
