Extracting safety information from multi-lingual accident reports using an ontology-based approach

Peter Hughes, Ryan Robinson, Miguel Figueres Esteban, Coen Van Gulijk

Research output: Contribution to journalArticle

Abstract

This paper describes an approach to extract meaning from multi-lingual free-text safety incident reports. A sample of 5065 safety incident reports from the Swiss Federal Office of Transport were used in the study. Each report was written in either German, French or Italian natural language. An interactive learning approach between a human and computer software was undertaken to identify key terms in the text that are relevant to discovering meaning. A multi-lingual ontology was created to join meaningful semantic patterns and identify specific classes of safety incident on the railway, including injuries occurring whilst passengers were boarding or alighting from vehicles, falling down stairs, struck by closing doors, or struck by objects such as suitcases. A graph database was used to query the text records via the ontology and identify reports of incidents in each class, regardless of the language used in the report. Fluent speakers of each language – German, French and Italian – reviewed the results to confirm true positive results and detect false positives. The performance of the process varied across languages and incident types, however the overall true positive rate was determined by the fluent speakers to be 98.5%.
LanguageEnglish
Pages288-297
Number of pages10
JournalSafety Science
Volume118
Early online date23 May 2019
DOIs
Publication statusE-pub ahead of print - 23 May 2019

Fingerprint

Tongue
ontology
Accidents
Ontology
incident
accident
Language
Safety
Stairs
Accidental Falls
language
Semantics
German language
Swiss
Software
Databases
German Federal Railways
semantics
Wounds and Injuries
learning

Cite this

@article{3d443bf73a2241b3b77ffbba02fe41c3,
title = "Extracting safety information from multi-lingual accident reports using an ontology-based approach",
abstract = "This paper describes an approach to extract meaning from multi-lingual free-text safety incident reports. A sample of 5065 safety incident reports from the Swiss Federal Office of Transport were used in the study. Each report was written in either German, French or Italian natural language. An interactive learning approach between a human and computer software was undertaken to identify key terms in the text that are relevant to discovering meaning. A multi-lingual ontology was created to join meaningful semantic patterns and identify specific classes of safety incident on the railway, including injuries occurring whilst passengers were boarding or alighting from vehicles, falling down stairs, struck by closing doors, or struck by objects such as suitcases. A graph database was used to query the text records via the ontology and identify reports of incidents in each class, regardless of the language used in the report. Fluent speakers of each language – German, French and Italian – reviewed the results to confirm true positive results and detect false positives. The performance of the process varied across languages and incident types, however the overall true positive rate was determined by the fluent speakers to be 98.5{\%}.",
keywords = "Accident analysis, safety ontology, natural language processing, railways",
author = "Peter Hughes and Ryan Robinson and {Figueres Esteban}, Miguel and {Van Gulijk}, Coen",
year = "2019",
month = "5",
day = "23",
doi = "10.1016/j.ssci.2019.05.029",
language = "English",
volume = "118",
pages = "288--297",
journal = "Safety Science",
issn = "0925-7535",
publisher = "Elsevier",

}

Extracting safety information from multi-lingual accident reports using an ontology-based approach. / Hughes, Peter; Robinson, Ryan; Figueres Esteban, Miguel; Van Gulijk, Coen.

In: Safety Science, Vol. 118, 01.10.2019, p. 288-297.

Research output: Contribution to journalArticle

TY - JOUR

T1 - Extracting safety information from multi-lingual accident reports using an ontology-based approach

AU - Hughes, Peter

AU - Robinson, Ryan

AU - Figueres Esteban, Miguel

AU - Van Gulijk, Coen

PY - 2019/5/23

Y1 - 2019/5/23

N2 - This paper describes an approach to extract meaning from multi-lingual free-text safety incident reports. A sample of 5065 safety incident reports from the Swiss Federal Office of Transport were used in the study. Each report was written in either German, French or Italian natural language. An interactive learning approach between a human and computer software was undertaken to identify key terms in the text that are relevant to discovering meaning. A multi-lingual ontology was created to join meaningful semantic patterns and identify specific classes of safety incident on the railway, including injuries occurring whilst passengers were boarding or alighting from vehicles, falling down stairs, struck by closing doors, or struck by objects such as suitcases. A graph database was used to query the text records via the ontology and identify reports of incidents in each class, regardless of the language used in the report. Fluent speakers of each language – German, French and Italian – reviewed the results to confirm true positive results and detect false positives. The performance of the process varied across languages and incident types, however the overall true positive rate was determined by the fluent speakers to be 98.5%.

AB - This paper describes an approach to extract meaning from multi-lingual free-text safety incident reports. A sample of 5065 safety incident reports from the Swiss Federal Office of Transport were used in the study. Each report was written in either German, French or Italian natural language. An interactive learning approach between a human and computer software was undertaken to identify key terms in the text that are relevant to discovering meaning. A multi-lingual ontology was created to join meaningful semantic patterns and identify specific classes of safety incident on the railway, including injuries occurring whilst passengers were boarding or alighting from vehicles, falling down stairs, struck by closing doors, or struck by objects such as suitcases. A graph database was used to query the text records via the ontology and identify reports of incidents in each class, regardless of the language used in the report. Fluent speakers of each language – German, French and Italian – reviewed the results to confirm true positive results and detect false positives. The performance of the process varied across languages and incident types, however the overall true positive rate was determined by the fluent speakers to be 98.5%.

KW - Accident analysis

KW - safety ontology

KW - natural language processing

KW - railways

U2 - 10.1016/j.ssci.2019.05.029

DO - 10.1016/j.ssci.2019.05.029

M3 - Article

VL - 118

SP - 288

EP - 297

JO - Safety Science

T2 - Safety Science

JF - Safety Science

SN - 0925-7535

ER -