Unifying Multisector Stakeholder Value Systems in Resilience Planning through Reinforcement Learning and Network Analysis
Conference
Ren, H, Zhang, L, Whetsell, TA et al. (2024). Unifying Multisector Stakeholder Value Systems in Resilience Planning through Reinforcement Learning and Network Analysis
. 508-517. 10.1061/9780784486139.055
Ren, H, Zhang, L, Whetsell, TA et al. (2024). Unifying Multisector Stakeholder Value Systems in Resilience Planning through Reinforcement Learning and Network Analysis
. 508-517. 10.1061/9780784486139.055
Resilience planning requires the integration of diverse stakeholder perspectives to formulate strategies that are effective and inclusive. Central to this process is understanding stakeholder value systems, defined as ranked systems of things that are of fundamental importance to stakeholders. Stakeholders from various sectors (e.g., public, private, and non-profit sectors) may have different value systems, which become the root cause of conflicts or disputes during the planning process. Thus, there is a critical need to systematically unite the varied value systems of stakeholders. To address this need, this study proposes an innovative approach that unifies the varying stakeholder value systems while considering the social influence of each stakeholder. This approach synergizes reinforcement learning, specifically Deep Q-learning, with network analysis. It begins with a network analysis that measures the social influence of stakeholders within a resilience planning collaboration network. It then introduces a novel reward function of the deep Q-learning algorithm to facilitate consensus in value systems among multiple stakeholders while ensuring minimal compromise from each stakeholder. Throughout the iterative process, the influence of each stakeholder in the network is considered. The proposed method is illustrated through a hypothetical case study focusing on uniting the value systems of diverse stakeholders in the context of housing resilience planning. The findings show that social influence has a significant impact on decision-making processes. This research provides a new framework for capturing and considering the complex dynamics of stakeholder interactions and value systems in resilience planning. The framework has the potential to enhance consensus-building in multisector stakeholder decision-making processes for resilience planning.
