Decoding the ‘Spirit of Place’: A Large Language Modeling (LLM) Approach to Formulating Perceived Pedestrian Accessibility in Space-Time Prisms

Abstract

Morphological discontinuities in urban fabric cause discrepancies between physical and
perceived accessibility. In response, this research explores how personal preferences and user
perception can be mathematically parameterized to bridge this gap, thereby improving the
representation of user accessibility in morphologically distinct urban environments. The study
proposes a novel methodology using large language modeling (LLM) to formulate how
phenomenological variables and "spirit of place" reshape pedestrian accessibility within space
time prisms (STPs) through AI-based simulations. A comparative analysis is conducted
between the morphologically continuous Amsterdam Centre Station Square and the spatially
fragmented Istanbul Beşiktaş Barbaros Square to decode decision-making strategies in these
contrasting environments. By utilizing LLMs to process qualitative preferences and narratives,
the research identifies how these experiences can be integrated as specific STP components.
This approach establishes a foundational AI-driven framework for bridging environmental
phenomenology with technical spatial modeling, creating a simulated environment for
subsequent quantitative applications and further research developed through real-user
interviews.