Author(s): Kai-Chi Huang; Gi-An Pen; Lin-Xiang Qiu; Po-Hsiang Chang; Su-Hui Chu; Po-Yin Chang

Linked Author(s): Po-Yin Chang

Keywords: Maritime training STCW certifications Large Language Models (LLMs) Retrieval-Augmented Generation (RAG) ChatGPT

Abstract: The rapid growth of the shipping industry has significantly increased the demand for skilled seafarers. However, the training process remains complex, requiring mastery of international maritime regulations and certifications mandated by the Standards of Training, Certification, and Watchkeeping for Seafarers (STCW) Convention. These certifications ensure that seafarers possess the necessary competencies to operate safely and effectively at sea but impose a heavy workload on both instructors and trainees. To address these challenges, this study leverages Large Language Models (LLMs) enhanced with Retrieval-Augmented Generation (RAG) to develop an AI assistant for maritime certification training. The AI assistant integrates external knowledge sources to provide precise, contextually relevant information through natural language interactions. Preliminary evaluations demonstrate that the RAG-based system delivers more accurate and beneficial responses compared to traditional LLMs like ChatGPT, particularly in domain-specific contexts. This project contributes to modernizing maritime education by improving learning efficiency and reducing information retrieval time for trainees. The findings offer valuable insights into integrating AI technologies into professional training programs within the maritime sector.

DOI:

Year: 2025