Uçar Kırmızıgül B., Demir B. K., Türkmen İ. U.
eCAADe 2025 Confluence, Ankara, Turkey, 1 - 05 September 2025, vol.1, pp.317-326, (Full Text)
-
Publication Type:
Conference Paper / Full Text
-
Volume:
1
-
City:
Ankara
-
Country:
Turkey
-
Page Numbers:
pp.317-326
-
TED University Affiliated:
Yes
Abstract
The rapid digitalization and widespread use of digital technologiesin the 21st centuryhave significantly transformed disciplines, includingurban studies. Today, understandingthe complex interplay between urban spaces and their usersrelies heavily on data-drivenmethods, necessitating innovative approaches to explore and enhance these interactions.This paper proposes a human-centric, data-driven interface that uses Large LanguageModels (LLMs) to analyze user-generated content. Unlike earlier methods such as LatentDirichlet Allocation (LDA), which struggledwith short, nuanced texts, this studyleverages advanced natural language processing (NLP) to offer deeper insights intourban experiences. The focus is on Kuğulu Park, a green space in Ankara, where userreviews are analyzed to understand behavioral,emotional, and thematic dimensions ofurban interactions. The interface integrates data collection,preprocessing, andvisualization processes, withadaptability for different datasets and analytical needs.Using advanced NLP models, itextracts thematic and sentiment-based insights even fromfragmented or context-rich user inputs.The study collects multilingual reviews andratings from publicly available platforms,offering a bottom-up, exploratory model forexamining urban behavior. By employing thisdata-driven approach, the paper aims toshed light on evolving perceptions of urban spaces and behaviors over time, correlatinguser sentiment with spatialand temporal dynamics. It underscores the transformativepotential of NLP models in urban studies,offering new opportunities for understandingand reshaping urban spaces through data-driven methodologies.