Researchers applied machine learning and natural language processing techniques to analyze aviation safety data from various sources. They used existing models and methods to uncover patterns contributing to incidents such as accidents and near-misses, and employed topic modelling to extract meaningful themes from unstructured reports. The study aimed to improve model transparency and trustworthiness by exploring causal inference techniques and interpretable AI frameworks. --- Why it matters: This research matters because it can provide valuable insights for aviation stakeholders, including regulators, airlines, and policymakers, to enhance incident analysis and decision making using data-driven solutions. Source: https://arxiv.org/abs/2608.14746