A Data-Driven Approach for Understanding Mobility Challenges at Airports

With the average age of the US population continuing to rise, demand for assisted mobility and accessible transport infrastructure is expected to grow. Airports have complex infrastructure and service environments that will need to adapt to cater to this changing demographic. It is therefore imperative to analyze the experience of navigating the complex airport infrastructure by people with limited mobility for future infrastructure planning. While conventional survey-based approaches have been well established to study user experiences, they are time-consuming, expensive, and small scale, restricting large-scale evaluation of accessibility experiences. This research presents a large language model (LLM)-based solution to examine the sentiments and experiences of passengers with disabilities or limited mobility while navigating through airport facilities. Drawing upon a large dataset of publicly available text from social media platforms such as Reddit, YouTube, and the Federal Aviation Administration complaint database, the LLM identifies patterns of satisfaction, frustration, and systemic barriers across different stages of the airport experience. By combining accessibility data with computational methods, this research demonstrates how LLMs can augment or partially automate traditional surveying processes, offering an efficient, low-cost, and context-aware approach to studying and analyzing user experience. The findings provide urban planners, airport authorities, and policymakers with actionable insights toward designing future airport infrastructure.