Risk factors extraction and analysis of Chinese ship collision accidents based on knowledge graph

Abstract

Shipping is a crucial mode of transportation. The high density of ship activities in Chinese waters increases the likelihood and severity of shipping accidents, which can significantly impact the global supply chain and shipping network operations. Among various maritime accidents, collisions are the most prevalent. Knowledge graphs, using triples (entity-relation-entity) as basic units, describe real-world concepts and relationships through text information, which aid in the causal analysis of accidents. This paper analyzes text data from Chinese ship collision accident reports and employs joint triple extraction algorithms based on deep learning and CART (Classification and Regression Tree) algorithm to construct a knowledge graph of these accidents, visualized using Gephi software. Utilizing complex network theory, a series of safety-related topological indicators are defined to perform quantitative risk assessment, identify key risk factors, and propose preventive measures, offering significant reference value for preventing ship collisions and other maritime accidents in Chinese waters.