An Integrated Framework of Association Rules and Social Network Analysis for Exploring Co-Occurrence Patterns of Waterborne Accident Causes

Accident cause analysis is widely valued by the International Maritime Organization (IMO) and many maritime authorities to prevent waterborne transportation accidents. In this study, 886 waterborne transportation accident investigation reports from China MSA are collected. Contributing factor analysis sections are extracted and imported into NVIVO for directed content coding, which combines the HFACS framework and grounded theory coding. In total, all contributing factors are identified, conceptualized, and categorized into 76 sub-categories across five dimensions. Based on these, association rule analysis is conducted separately for seven accident types to mine corresponding strong association rules. Meanwhile, an undirected co-occurrence social network is constructed with all accident samples and analyzed from three dimensions: connection strength, criticality, and activeness. Multi-dimensional network structure analysis shows that external influence accident causes do not rank among the top nodes across all indicators. Within the scope of this study, the contributing factor co-occurrence network presents a closed-like structural feature dominated by internal factors. “Failure to identify risks timely and take effective measures” and “Incompetent crew” are identified as two core nodes with outstanding comprehensive performance across multiple indicators. In addition, nodes with prominent local co-occurrence aggregation characteristics are also identified through clustering coefficient analysis.