Text mining techniques to identify causes and hazards of ship fire accidents
0
조회
0
다운로드
0
인용
0
참고문헌
키워드
초록
SOLAS regulation II-2/17 (MSC/Circ. 1002) allows the use of performance-based fire engineering design to demonstrate that design solutions not complying with traditional prescriptive-based regulations are as safe as equivalent design solutions. Howev-er, in the field of fire simulation, which constitutes a large part of performance-based design, there is a paucity of data to set a de-tailed design of a ship fire scenario. Therefore, the importance of fire hazard elements in designing fire scenarios can vary based on the user’s judgments. The purpose of the study is to identify the topics of the causes and hazard of fire accidents for use in the design of fire scenarios. Hence, we used topic modeling of text mining techniques (a text data analysis method) to analyze the written ver-dict of fire accidents over the past decade. The analysis results are summarized as follows. First, the result of topic modeling indicat-ed that fires due to electrical, mechanical, and carelessness accounted for 69.4%, 17%, and 12.2%, respectively, of the total analyzed fires. Second, fire hazard keywords as per the cause of fire were identified based on the fire cause group obtained through topic mod-eling. Electrical fire hazard keywords with the highest frequency include “engine room” for fire occurrence location, “electric wire” for flammable material, a “short circuit” for ignition factor, and “flame” for an ignition source.
Abstract
SOLAS regulation II-2/17 (MSC/Circ. 1002) allows the use of performance-based fire engineering design to demonstrate that design solutions not complying with traditional prescriptive-based regulations are as safe as equivalent design solutions. Howev-er, in the field of fire simulation, which constitutes a large part of performance-based design, there is a paucity of data to set a de-tailed design of a ship fire scenario. Therefore, the importance of fire hazard elements in designing fire scenarios can vary based on the user’s judgments. The purpose of the study is to identify the topics of the causes and hazard of fire accidents for use in the design of fire scenarios. Hence, we used topic modeling of text mining techniques (a text data analysis method) to analyze the written ver-dict of fire accidents over the past decade. The analysis results are summarized as follows. First, the result of topic modeling indicat-ed that fires due to electrical, mechanical, and carelessness accounted for 69.4%, 17%, and 12.2%, respectively, of the total analyzed fires. Second, fire hazard keywords as per the cause of fire were identified based on the fire cause group obtained through topic mod-eling. Electrical fire hazard keywords with the highest frequency include “engine room” for fire occurrence location, “electric wire” for flammable material, a “short circuit” for ignition factor, and “flame” for an ignition source.
참고문헌 (16)
- 황광일, "선상안전교육 시 선내 연기충진 여부에 따른 실습생의 피난이동특성 비교", 해양환경안전학회지, 24(4), pp.422~429 (2018), 해양환경안전학회[DOI]
- 김별, "인명안전 관점에서 선박 화재 시 급·배기조건에 따른 배연성능 예측평가", 해양환경안전학회지, 22(7), pp.782~790 (2016), 해양환경안전학회[DOI]
- H. J. Kang, "A frame-work for using computational fire simulations in the early phases of ship design", Ocean Engineering, 129, pp.335~342 (2017)[DOI]
- International Maritime Organization (IMO), "Guidelines on alternative design and arrangements for fire safety, UK, MSC", Circ. 1002, 2001
- "Statistics for the marine accidents"[URL]
- "Written verdicts of Korea maritime safety tribunal"[URL]
- 김범수, "사고보고문서를 이용한 텍스트 기반 사고발생 유형 및 관계 분석", 한국안전학회지, 33(3), pp.58~64 (2018), 한국안전학회[DOI]
- 강성식, "재해분석을 위한 텍스트마이닝과 SOM 기반 위험요인지도 개발", 한국안전학회지, 33(6), pp.77~84 (2018), 한국안전학회[DOI]
- 강석용, "빅데이터 기법을 활용한 항해 중 준해양사고 발생원인 분석에 관한 연구", 해양환경안전학회지, 24(4), pp.408~414 (2018), 해양환경안전학회[DOI]
- A. Hotho, "A brief survey of text mining", Journal for Computational Linguistics and Language Technology, 20(1), pp.19~62 (2005)
- Y. M. Baek, "Text Mining Using R" (2017), Han-ul Acad-emy
- D. M. Blei, "Latent dirichlet allocation", Journal of Machine Learning Research, 3, pp.993~1022 (2003)
- 김민관, "토픽 모델링을 활용한 컨설팅 연구동향 분석", 한국산업경영시스템학회지, 40(4), pp.46~54 (2017), 한국산업경영시스템학회
- D. M. Blei, "Probabilistic topic models", Communica-tions of the ACM, 55(4), pp.77~84 (2012)
- 박자현, "토픽모델링을 활용한 국내 문헌정보학 연구동향 분석", 정보관리학회지, 30(1), pp.7~32 (2013), 한국정보관리학회[DOI]
- B. Grün, "Topicsmodels : An R package for fitting topic models", Journal of Statistical Software, 40(13), pp.1~30 (2011)