Abstract
In recent years, socio-technical systems, including aviation, require “resilience,” the ability of people to maintain or improve their functioning by responding and adjusting flexibly to changing circumstances. To enhance resilience, focusing on daily successes (normal operations) and learning from them is necessary. However, it is not easy to learn from normal operations to understand the factors that led to why they succeeded.
Therefore, attempts have been made to analyze resilience by regarding near-miss events as “successful cases in which accidents were avoided due to flexible human actions.” Since analyzing the large volume of near-miss events collected daily requires significant time and cost, attention has been drawn to the utilization of natural language processing.
In this study, we used near-miss event data from the Aviation Safety Reporting System to extract actions that demonstrated resilience. Furthermore, we attempted an analysis using GPT, a natural language processing technique, evaluated the validity of the responses obtained, and explored methods to enhance their validity.
Access this chapter
Tax calculation will be finalised at checkout
Purchases are for personal use only