Abstract
PurposeMegaproject knowledge innovation (MKI) is perceived as a critical strategy for engineering value co-creation and industrial chain upgrading. Ascertaining the impact mechanism of MKI is a crucial initial step towards improving management practices. Within the framework of complex systems in megaprojects, factors exhibit intricate interdependencies. However, the current domain of knowledge has either overlooked or oversimplified this relationship and therefore cannot propose pragmatic and efficacious strategies for enhancing MKI. To close this gap, this study develops a Bayesian network (BN) model aiming to investigate the interdependencies among MKI-related factors and their impact on MKI.Design/methodology/approachFirst, this study implements literature review, expert interview and field investigation to identify the influencing factor nodes for the network model development. Second, a Bayesian network was constructed by integrating the expert knowledge with Dempster-Shafer theory. Next, a MKI measurement model was established using 253 training samples. Finally, the factor significance and optimal MKI improvement strategies are identified from the sensitivity analysis and probabilistic reasoning within the BNs.FindingsThe results indicate that (1) the BN model exhibits significant reliability and holds promotion and application value in formulating MKI management strategies; (2) knowledge sharing, shared vision and leadership are the key influencing factors of MKI; and (3) simultaneously improving institutional pressure, leadership and knowledge sharing is the most optimal strategy to enhance MKI.Originality/valueThis study innovatively introduced the BN method into the domain of MKI management, providing an appropriate approach for modelling complex relationships among factors and investigate nonlinear influences. The developed model raises megaproject stakeholders’ awareness about factors influencing MKI and presents quantified strategies that increase the likelihood of maximising MKI levels. Its ease of generalisability positions it as a promising decision support tool, facilitating the implementation of sustainable MKI practices.