Special session: Reading the City: AI-driven Sentiment Analysis for Urban Decision-making - A Methodological Workshop on Integrating Resident Sentiment into Evidence-based Urban Planning

Urban planners increasingly face a fundamental disconnect: while cities generate unprecedented volumes of spatial and socioeconomic data, understanding how residents perceive and experience urban environments remains methodologically challenging. This session examines how artificial intelligence can systematically analyse resident sentiment and correlate it with traditional planning indicators, offering new insights into the relationship between measured urban performance and lived experience - true human centricity. The session centres on a case study using GUS, an AI-powered platform that applies natural language processing, computer vision, and retrieval-augmented generation to analyse resident feedback from multiple sources like social media, surveys, and public commentary. This perception data is then spatially mapped and cross-referenced with established indicators including housing accessibility, employment density, transport coverage, and urban vitality indices. The methodology reveals patterns and disparities that conventional analysis often overlooks. The presentation follows a structured format. Beginning with an overview of AI applications in perception analysis for planning contexts, the session then moves to hands-on exploration of platform outputs including sentiment-satisfaction mapping, temporal topic clustering, and multi-factor correlation matrices. Participants will examine specific cases where perception data diverged significantly from quantitative indicators, for instance, employment-dense areas reporting low opportunity perception, or well-resourced neighbourhoods showing unexpected dissatisfaction patterns. A key component involves participant interaction with the data, allowing attendees to explore correlations, test assumptions, and critically evaluate the methodology. The session addresses both technical considerations 1) how AI processes and categorises qualitative data and 2) practical implications for planning practice. The discussion component examines broader questions about integrating AI-derived perception analysis into planning frameworks. What validation methods ensure reliability? How should planners weight sentiment data against established metrics? What ethical considerations arise when using AI to interpret community voice? This session contributes to ongoing discourse about evidence-based planning by presenting a reproducible methodology for incorporating resident perception at scale. The approach demonstrated offers one model for addressing the persistent challenge of connecting quantitative urban performance metrics with qualitative resident experience, particularly relevant for equity-focused planning initiatives. Intended for researchers, practitioners, and policymakers engaged with smart city initiatives, participatory planning methods, and the integration of emerging technologies in urban analysis.

This session explores how AI can analyse resident sentiment from social media, surveys, and public comments, then map these insights against urban indicators like housing and employment access. Using the GUS platform, participants will examine real cases where perception data contradicts conventional metrics, discovering new methods to understand the gap between measured urban performance and actual resident experience.