Exploring Quantum Natural Language Processing in Risk Analysis: A Comparative Study on Aviation Accident Reports

Quantum computing, leveraging properties like superposition and entanglement, has transformative potential across diverse domains. However, Quantum Natural Language Processing (QNLP) applications in Risk Analysis (RA), particularly in aviation, remain understudied. This research bridges this gap by comparing QNLP models to classify aviation accidents, focusing on human error. Using Lambeq’s pipeline to convert sentences into quantum circuits, we analyze accident narratives from the National Transportation Safety Board (NTSB) database. Our study tests three models: two quantum and one hybrid. The hybrid model, achieving 80.48% accuracy, outperforms purely quantum models and a classical TF-IDF NLP model in nuanced semantic relationships. This research highlights QNLP’s potential to enhance accuracy and efficiency in accident classification, with future work aimed at validating these models on real quantum hardware.