Accidents in aviation are rare events. From them, aviation safety management systems take fast and effective remedy actions by performing the analysis of the root causes of accidents, most of which are proved to be human factors. Since the current standard relies on the manual classification performed by trained staff, there are no technical standards already defined for automated human factors identification. This paper considers this issue, proposing machine learning techniques by leveraging on the state-of-the-art technologies of Natural Language Processing. The techniques are then adapted to the Software Hardware Environment Liveware (SHEL) standard accident causality model and tested on a set of real accidents. The computational results show the accuracy and effectiveness of the proposed methodology. Furthermore, the application of the methodology to real documents checked by experts estimates a reduction of the time needed for at least 30% compared to the standard methods of human factors identification.

Natural Language Processing for the identification of Human factors in aviation accidents causes: An application to the SHEL methodology / Perboli, Guido; Gajetti, Marco; Fedorov, Stanislav; Giudice, Simona Lo. - In: EXPERT SYSTEMS WITH APPLICATIONS. - ISSN 0957-4174. - STAMPA. - 186:(2021), p. 115694. [10.1016/j.eswa.2021.115694]

Natural Language Processing for the identification of Human factors in aviation accidents causes: An application to the SHEL methodology

Perboli, Guido;Fedorov, Stanislav;
2021

Abstract

Accidents in aviation are rare events. From them, aviation safety management systems take fast and effective remedy actions by performing the analysis of the root causes of accidents, most of which are proved to be human factors. Since the current standard relies on the manual classification performed by trained staff, there are no technical standards already defined for automated human factors identification. This paper considers this issue, proposing machine learning techniques by leveraging on the state-of-the-art technologies of Natural Language Processing. The techniques are then adapted to the Software Hardware Environment Liveware (SHEL) standard accident causality model and tested on a set of real accidents. The computational results show the accuracy and effectiveness of the proposed methodology. Furthermore, the application of the methodology to real documents checked by experts estimates a reduction of the time needed for at least 30% compared to the standard methods of human factors identification.
File in questo prodotto:
File Dimensione Formato  
2021-ESWA-Natural Language Processing for the identification of Human factors in aviation accidents .pdf

non disponibili

Descrizione: Post Print version
Tipologia: 2a Post-print versione editoriale / Version of Record
Licenza: Non Pubblico - Accesso privato/ristretto
Dimensione 769.53 kB
Formato Adobe PDF
769.53 kB Adobe PDF   Visualizza/Apri   Richiedi una copia
main.pdf

accesso aperto

Descrizione: Pre Print version
Tipologia: 1. Preprint / submitted version [pre- review]
Licenza: PUBBLICO - Tutti i diritti riservati
Dimensione 343.68 kB
Formato Adobe PDF
343.68 kB Adobe PDF Visualizza/Apri
Pubblicazioni consigliate

I documenti in IRIS sono protetti da copyright e tutti i diritti sono riservati, salvo diversa indicazione.

Utilizza questo identificativo per citare o creare un link a questo documento: https://hdl.handle.net/11583/2917532