AI-Driven Governance: Intelligent Stakeholder Strategy for Mega-infrastructure Projects

Authors

• Jin Xue, University of SydneyFollow
• Jacky Wei-Ting Hong, University of SydneyFollow
• Qian-Cheng Wang, City University of Hong KongFollow
• Yunping Liang, University of Nebraska-LincolnFollow
• Jennifer Whyte, University of SydneyFollow
• Kerui Lyu, University of SydneyFollow

Document Type

Report

Date of this Version

5-13-2026

Citation

APM Research Fund Series, Association for Project Management (APM), Princes Risborough, UK. <https://www.apm.org.uk/resources/research/published-research/ai-driven-governance-intelligent-stakeholder-strategy-for-mega-infrastructure-projects/>, May 13, 2026

Abstract

Executive summary 1.1 Project overview Mega-infrastructure projects operate in volatile environments where stakeholder interests change rapidly. Management of these relationships, particularly regarding environmental, social and governance (ESG) issues, is often fragmented and difficult to track. The objective of this project is to develop an AI-driven systematic framework to enhance stakeholder management by understanding the complex and dynamic nature of stakeholder engagement and deriving actionable insights for project managers. In achieving this, the project delivered an innovative solution that leverages artificial intelligence (AI) and large language models (LLMs) to analyse official engagement records from case projects across three countries, which enabled stakeholder entity and stakeholder-event extraction. The core outputs in this framework include: • a strategic stakeholder engagement map to illustrate the alignment between intended and actual engagement, revealing scenarios of proportionate engagement, over-engagement or under-engagement • data-driven knowledge graphs (KGs) as visual networks that link stakeholders directly to specific events • an LLM-powered chatbot that answers issue-level questions using KGs and source texts • a web-based platform designed to enhance stakeholder management practices for industrial practitioners, which integrates: • Knowledge Base, where users upload and manage project document sets and trigger KG construction • an interactive KG visualiser that renders stakeholders (organisations) and events (documented actions or decisions) as a network • an LLM-powered chatbot that answers issue-level questions using KGs and source texts. 1.2 Key findings To evaluate the models developed, three megaproject case studies were selected, and their publicly available reports were used as data input. Within these cases, the strategic stakeholder engagement map has shown different insightful results where alignment, over-engagement and under-engagement scenarios can all be observed. In parallel, KGs are constructed with two node types – stakeholders and events – and event-mediated edges enabling centrality-based filtering and phase snapshots. 1.3 Recommendations This research bridges the gap between academic theory and industrial practice by offering a tangible web-based stakeholder knowledge platform. For corporate partners and project practitioners, this platform could be used for the following: • Monitor the stakeholder engagement strategy execution gap: Use the strategic engagement map to monitor the awareness–outcome gap in an existing strategy. Shift resources away from ‘overengaged’ areas to address ‘under-engaged’ risks. • Monitoring dynamic networks: Move beyond static stakeholder registers. Use KGs to visualise how stakeholder relationships evolve over time. • Manage the project dataset: Use the platform as a knowledge management tool to upload and break down historical project reports into structured, reusable evidence for future analysis. • Utilising the web platform: Upload project records and instantly generate visual insights. Ask the AI chatbot specific questions (e.g. ‘Who raised environmental concerns in Phase 1?’), and review answers linked directly to the original documents for verification.