Illuminating themes of ecovillages by leveraging participatory modeling: A convergent Gen-AI and Fuzzy Cognitive Maps approach.

Ecovillages, as defined by the United Nations, are communities that focus primarily on aspects of sustainability. The research on ecovillages is vital, as it helps to uncover actionable artefacts of sustainable practices. However, much of the extant literature has examined individual ecovillages; thereby leading us to investigate broader themes of ecovillages holistically. This data-driven study aimed to uncover themes of ecovillages and attempted to furnish a comprehensive ecovillage model for quantitative assessment. We compiled ecovillage descriptions from the Global Ecovillage Network (GEN) to create the Comprehensive Ecovillage Description Dataset (CEDD). A generative AI model, GPT-4o, was then used to semantically extract key terms from CEDD, and the results were further processed using the topic modeling technique BERTopic to identify prominent themes. These themes were shared with 18 stakeholders to undertake the development of a Unified Ecovillage Map (UEM), using Participatory Modeling (PM) and Fuzzy Cognitive Maps (FCMs). To understand these themes, we conducted scenario analyses of UEM using Mental Modeler. This structured "what-if" simulation approach, allowed us to assess the potential impact of various sustainability initiatives. This data-driven scenario analyses revealed numerous insights-for instance, initiatives such as permaculture design and renewable energy adoption enhance ecological balance and energy optimization, while measures like incoming visitors may bolster financial liquidity but simultaneously challenge cultural preservation and communal harmony. This research seeks to offer a nuanced, stakeholder-centric approach by examining ecovillage themes which can effectively delineate managerial recommendations and policy implications for enhancing ecovillage management.