Integrating BERT and DIM–SD: a hybrid model for simulating operational performance of education infrastructure PPP projects

Purpose This study aims to construct an integrated Bidirectional Encoder Representations from Transformers (BERT) and DEMATEL–ISM–MICMAC (DIM)–system dynamics (SD) operational performance simulation model for the education Public Private Partnership projects (PPPs). The model can simulate the implementation effects of operational strategies, providing practical support for decision-making. This study contributes to addressing the issue of decision failure caused by the dynamic and uncertain nature of operations, helping project companies in education PPPs to achieve the performance objectives in operational contracts steadily. Design/methodology/approach First, a BERT-based natural language processing (NLP) model is developed to automatically identify operational performance indicators from PPP contracts, establishing a performance evaluation indicator system. Second, a DIM model is adopted, in which the Delphi method is used to score the influence relationships among the indicators, constructing a structural model of the operational performance system. Finally, based on the performance indicators and the system structure model, an SD simulation model is established. Findings To verify the effectiveness of the model, a newly operated university PPP project in Yantai, China, is used as a case study. The results demonstrate that the simulated performance is consistent with the actual operational performance. The model can simulate the dynamic changes of operational system and test the implementation effects of operational strategies, providing support for operational decision-making in project companies. Originality/value This study proposes an integrated BERT–DIM–SD performance simulation model framework. First, applying NLP technology overcomes the limitations of manually identifying performance indicators. Second, the DIM method is employed to construct a structured model of the operational performance indicator system, thereby enriching the theoretical foundation of performance management in education PPPs. Furthermore, integrating DIM with SD establishes a feasible pathway for translating DIM's theoretical analysis results into operational practice. Finally, the integrated simulation model can simulate the dynamic changes of education PPPs operations and test the effectiveness of different strategies, offering project companies a practical decision-support tool to address decision failure issues.