The engineering–procurement–construction (EPC) model has been widely adopted in construction practice for its ability to achieve efficient integration and unified coordination across project phases. However, the diversity and complexity of cost risk factors in EPC projects pose significant challenges to effective cost control. This study systematically analyzes 212 judicial cases from the China Judgments Online database to extract and categorize 24 cost risk factors, forming a multilevel risk structure. Social network analysis (SNA) is employed to identify the structural relationships among these risk factors, while the Apriori algorithm is applied to uncover frequent cooccurrence patterns, revealing the coupling mechanisms and transmission pathways of core risks. The results indicate that a system science–based classification approach effectively supports the identification and modeling of complex risk networks. Legal, technical, and managerial risks emerge as key nodes contributing to cost overruns, with strong coupling and bidirectional feedback observed among multiple factors. Furthermore, the proposed data-driven analytical framework demonstrates strong cross-jurisdictional applicability, providing theoretical and methodological guidance for project risk governance under different legal and regulatory contexts. This research deepens the theoretical understanding of systemic cost risk behaviors in EPC projects and offers empirical and practical insights for improving contract governance and risk control in international EPC practices.
Wang et al. (2026) studied this question.