Enhancing collaborative resilience planning

Unifying stakeholder value systems through integrating reinforcement learning and network analysis

Bibliographic Data

ID6418973
AuthorsHang Ren (0000-0001-5326-6433, Virginia Tech), Lu Zhang (0000-0002-8855-8492, Virginia Tech, corresponding author), Travis A Whetsell (0000-0001-5395-4754, Georgia Institute of Technology), N Emel Ganapati (0000-0003-0967-9593)
Year2025
Volume170
Pages106635-106635
Publication date2025-11-12
Open AccessYes
TypeARTICLE
VenueCities (JOURNAL)
Journal identifiersISSN: 0264-2751 • E-ISSN: 1873-6084
PublisherElsevier BV (PUBLISHER)
DOI10.1016/j.cities.2025.106635
OpenAlexW7105515805
LanguageEN
References cited72

Resilience planning necessitates the integration of diverse stakeholder values – the things that matter most to stakeholders – to formulate strategies that are effective and inclusive. However, stakeholders from different sectors (e.g., public, private, non-profit) hold numerous values with varying degrees of importance, forming their unique value systems. The divergent value systems can become a root cause of conflicts and impede consensus reaching. Currently, there is limited research on how to unify diverse stakeholder value systems to resolve conflicts in resilience planning. To address this methodological gap, this paper introduces a new Stakeholder Value Unification (SVU) model, which combines network analysis with reinforcement learning to support consensus-building in resilience planning. The SVU model uses stakeholder survey data on value priorities and collaboration networks as inputs. It first assesses each stakeholder's structural positions through centrality measures, then incorporates these positions into a deep Q-learning algorithm. By framing consensus-building as a sequential decision-making problem, the reinforcement learning component identifies value adjustment strategies that maximize collective agreement while minimizing individual concessions. The SVU model offers a systematic approach to reconcile divergent stakeholder value systems while accounting for their network positions to foster more collaborative planning outcomes. We demonstrate the applicability of the proposed SVU model under multiple scenarios through hypothetical and real-world case studies on housing resilience planning. The model provides policymakers with a tool for systematically incorporating value systems of different stakeholders into resilience planning, thus potentially supporting multisector stakeholder collaboration and facilitating the development of more socially acceptable resilience strategies.

Centrality · Learning network · Network analysis · Reinforcement learning · Resilience (materials science · Social network analysis · Stakeholder · Stakeholder analysis · Unification · Value (mathematics · Value network · Collaboration in agile enterprises · Housing, Finance, and Neoliberalism · Infrastructure Resilience and Vulnerability Analysis · Supply Chain Resilience and Risk Management · Sustainability and Climate Change Governance · Urban Planning and Governance

Tools

Open DOIOpen Access