Nurduhan M., Kuleyin B.

8. Uluslararası Asya Modern Bilimler Kongresi, Aksaray, Turkey, 5 - 07 May 2023, pp.525

  • Publication Type: Conference Paper / Summary Text
  • City: Aksaray
  • Country: Turkey
  • Page Numbers: pp.525
  • Dokuz Eylül University Affiliated: Yes


In studies conducted on the causes of maritime accidents, it has been determined that 70-90% of accidents occur as a result of "human error". Therefore, the increasing number of maritime accident studies, the number of which has increased since the 1970s, started to be carried out mostly on the basis of "human error" after 2009. This study aims to identify research trends related to human errors in maritime accidents. To achieve this aim, studies focusing on the concept of human error in maritime accidents were comprehensively analysed using topic modelling methods. In the study, Abstracts and publication years of 339 articles in the SCOPUS database were used as the dataset. Within the scope of the analyses conducted in the study, the prominent keywords in the examined studies were determined as "collision", "grounding", "HFACS", "communication", and "equipment". In addition, it was observed that the studies in question were divided into three clusters called "method", "cause of accident", and "collision" by clustering algorithm. However, it was determined that the keywords "cruise", "tanker", "Vessel Traffic Services (VTS)", and "collision risk" stood out in the studies between 2020-2023 and the studies focused particularly on tanker and passenger ships. When examining the main keywords that stand out in the studies by year, it is seen that the words "safety culture", "sleep", and "fatigue", which are frequently encountered in maritime accidents in recent years, can be successfully identified. In this study, since the TF-IDF algorithms used to determine keywords overlook the semantic similarities between sentences, topic modelling methods based on semantic similarity can be used in future studies. In addition, studies on human errors in accidents occurring in critical industries such as aviation and nuclear facilities can also be analysed through topic modelling to examine similarities and/or differences.