Bibliometric analysis of aviation unsafe event research: from reactive response to intelligent prediction
Fuzheng Zhang, Xiaomei Zhang, Yongxin GuanPurpose
The post-pandemic recovery of global air transport and rapid developments in artificial intelligence are shifting aviation safety management from reactive accident investigation to proactive risk prediction. To examine how this trend is reflected in recent literature, this study aims to conduct a bibliometric analysis of aviation unsafe event research published between 2020 and 2025.
Design/methodology/approach
Based on 512 publications retrieved from the Web of Science Core Collection, the authors use Bibliometrix and VOSviewer to construct keyword co-occurrence and co-citation networks.
Findings
The results show a rapid increase in research output following a temporary decline in 2022, reflecting the field’s resilience and growing attention to post-pandemic safety challenges. Thematic evolution analysis identifies a clear departure from traditional system-level investigations. This field increasingly prioritizes mining valuable information from flight data and unstructured safety reports using machine learning, deep learning and natural language processing. Current studies largely revolve around anomaly detection under complex meteorological conditions, causal inference using Bayesian networks and the application of large language models for safety text analysis.
Originality/value
These findings suggest that data-driven approaches have established themselves in aviation unsafe event research. Future studies should focus on addressing the interpretability of deep learning models and optimizing human–machine collaboration in safety-critical systems.