การประยุกต์ใช้ปัญญาประดิษฐ์ในการสืบค้นและสรุปข้อมูลเมือง

Jula Jiratrakanvong; Nattawut Preyawanit

Authors

Jula Jiratrakanvong Faculty of Architecture, Silpakorn University
Nattawut Preyawanit Faculty of Architecture, Silpakorn University

Keywords:

Artificial intelligence, Urban Data, Open Data

Abstract

This research article aimed to 1. analyze the characteristics, structure, and limitations of urban datasets used for designing a prototype City Intelligence Platform, 2. design and develop a prototype system that applies artificial intelligence to integrate, retrieve, analyze, and summarize urban data, and 3. evaluate the capabilities and limitations of the prototype system in interpreting user queries, selecting appropriate data, analyzing information, and generating traceable outputs. This study employed a research and development approach. Urban datasets from Bangkok and Chiang Mai were used as the data sources, including air quality, traffic accident, traffic, and spatial data obtained from publicly available databases. The research instruments consisted of the prototype system, artificial intelligence prompts, data analysis tools, and a set of 22 evaluation questions. Data were collected by preparing datasets according to the predefined study areas, time periods, and variables, followed by experiments using evaluation questions covering the spatial, temporal, and attribute dimensions of urban data. Data were analyzed by evaluating the accuracy of interpreted parameters, the appropriateness of the selected data and analytical tools, and the correctness of numerical and textual responses. Absolute error, mean absolute percentage error, and descriptive analysis were employed for data analysis.

The research findings revealed that 1. urban datasets from different sources varied in their structures, accessibility, levels of detail, and metadata, requiring specific data preparation processes and dedicated analytical tools for each data type when integrating information across multiple sources; 2. the prototype system was capable of functioning as an intermediary between natural language queries and urban datasets from multiple sources by employing an agentic artificial intelligence mechanism together with a retrieval-based answer generation approach, enabling the generation of traceable outputs accompanied by supporting information; and 3. based on the evaluation using 22 test questions, the system completely delivered the required information in 20 cases and partially delivered it in 2 cases. Among 79 interpreted parameters, 78 were completely correct. Numerical and date/time responses achieved a mean absolute percentage error of 0.04%, whereas 3 out of 5 responses requiring textual reasoning or evidence-based interpretation were incorrect. These findings indicate that the proposed system has the potential to support urban data access and analysis, particularly for tasks involving well-structured data. However, practical implementation should incorporate different levels of human verification according to the characteristics of the generated outputs and should present data sources, parameters, metadata, and supporting evidence together with the responses to enhance system reliability.

References

Cheung, J. C. & Ho, S. S. (2025). The effectiveness of explainable AI on human factors in trust models. Scientific Reports, 15, 23337.

Del Rosario, R. F. et al. (2025). Architecting resilient LLM agents: A guide to secure plan-then-execute implementations. arXiv preprint arXiv:2509.08646.

Gantla, H. et al. (2025). Fusion of real-time traffic and environmental sensor data with machine learning for optimizing smart city operations. Fusion: Practice and Applications, 19(2), 328–340.

He, M. et al. (2025). Generative AI for urban design: A stepwise approach integrating human expertise with multimodal diffusion models. arXiv preprint arXiv:2505.24260.

Kanna, K. & Kolbe, T. H. (2025). Advanced user interaction with urban digital twins using large language models. ISPRS Annals of the Photogrammetry, Remote Sensing and Spatial Information Sciences, X-G-2025, 469–476.

Kaynak, S. et al. (2025). City-scale digital twin framework for flood impact analysis: Integrating urban infrastructure and real-time data analytics. Urban Climate, 64, 102640.

Lin, T. (2025). Research on cross-domain data fusion and co-optimization strategy of smart city management software. Frontiers in Science and Engineering, 5(7), 35–46.

Padsala, R. et al. (2025). A data-driven urban digital twin approach for evaluating positive energy district potential using OGC standards in Stuttgart. The International Archives of the Photogrammetry, Remote Sensing and Spatial Information Sciences, XLVIII-4/W16-2025, 67–74.

Pan, Y. et al. (2016). Urban big data and the development of city intelligence. Engineering, 2(2), 171–178.

Tiwari, A. (2025). Conceptualising the emergence of agentic urban AI: From automation to agency. Urban Informatics, 4, 1–16.

Wilson, E. et al. (2025). Real-time data visualization frameworks for smart city applications [Preprint]. ResearchGate. doi:10.13140/RG.2.2.16704.85762.