Logo
ScienceStack
Table of Contents
Fetching ...
Sign In

Urban-MAS: Human-Centered Urban Prediction with LLM-Based Multi-Agent System

Shangyu Lou

TL;DR

This work tackles the challenge of applying LLMs to human-centered urban prediction by introducing Urban-MAS, a three-layer, zero-shot multi-agent system. It combines Predictive Factor Guidance, Reliable UrbanInfo Extraction, and Multi-UrbanInfo Inference to prioritize influential factors, ensure reliable information extraction from multi-source data, and fuse signals across social and built environments at macro and street scales. Across urban perception and running amount tasks in Tokyo, Milan, and Seattle, Urban-MAS yields significant reductions in MAE, MSE, and RMSE compared with a single LLM baseline, with ablations highlighting the critical role of predictive factor prioritization and robust extraction. The approach offers a scalable, transferable paradigm for human-centered urban AI that supports cross-city generalization and robust decision support for urban policymaking.

Abstract

Urban Artificial Intelligence (Urban AI) has advanced human-centered urban tasks such as perception prediction and human dynamics. Large Language Models (LLMs) can integrate multimodal inputs to address heterogeneous data in complex urban systems but often underperform on domain-specific tasks. Urban-MAS, an LLM-based Multi-Agent System (MAS) framework, is introduced for human-centered urban prediction under zero-shot settings. It includes three agent types: Predictive Factor Guidance Agents, which prioritize key predictive factors to guide knowledge extraction and enhance the effectiveness of compressed urban knowledge in LLMs; Reliable UrbanInfo Extraction Agents, which improve robustness by comparing multiple outputs, validating consistency, and re-extracting when conflicts occur; and Multi-UrbanInfo Inference Agents, which integrate extracted multi-source information across dimensions for prediction. Experiments on running-amount prediction and urban perception across Tokyo, Milan, and Seattle demonstrate that Urban-MAS substantially reduces errors compared to single-LLM baselines. Ablation studies indicate that Predictive Factor Guidance Agents are most critical for enhancing predictive performance, positioning Urban-MAS as a scalable paradigm for human-centered urban AI prediction. Code is available on the project website:https://github.com/THETUREHOOHA/UrbanMAS

Urban-MAS: Human-Centered Urban Prediction with LLM-Based Multi-Agent System

TL;DR

This work tackles the challenge of applying LLMs to human-centered urban prediction by introducing Urban-MAS, a three-layer, zero-shot multi-agent system. It combines Predictive Factor Guidance, Reliable UrbanInfo Extraction, and Multi-UrbanInfo Inference to prioritize influential factors, ensure reliable information extraction from multi-source data, and fuse signals across social and built environments at macro and street scales. Across urban perception and running amount tasks in Tokyo, Milan, and Seattle, Urban-MAS yields significant reductions in MAE, MSE, and RMSE compared with a single LLM baseline, with ablations highlighting the critical role of predictive factor prioritization and robust extraction. The approach offers a scalable, transferable paradigm for human-centered urban AI that supports cross-city generalization and robust decision support for urban policymaking.

Abstract

Urban Artificial Intelligence (Urban AI) has advanced human-centered urban tasks such as perception prediction and human dynamics. Large Language Models (LLMs) can integrate multimodal inputs to address heterogeneous data in complex urban systems but often underperform on domain-specific tasks. Urban-MAS, an LLM-based Multi-Agent System (MAS) framework, is introduced for human-centered urban prediction under zero-shot settings. It includes three agent types: Predictive Factor Guidance Agents, which prioritize key predictive factors to guide knowledge extraction and enhance the effectiveness of compressed urban knowledge in LLMs; Reliable UrbanInfo Extraction Agents, which improve robustness by comparing multiple outputs, validating consistency, and re-extracting when conflicts occur; and Multi-UrbanInfo Inference Agents, which integrate extracted multi-source information across dimensions for prediction. Experiments on running-amount prediction and urban perception across Tokyo, Milan, and Seattle demonstrate that Urban-MAS substantially reduces errors compared to single-LLM baselines. Ablation studies indicate that Predictive Factor Guidance Agents are most critical for enhancing predictive performance, positioning Urban-MAS as a scalable paradigm for human-centered urban AI prediction. Code is available on the project website:https://github.com/THETUREHOOHA/UrbanMAS

Paper Structure

This paper contains 7 sections, 1 equation, 1 figure, 2 tables.

Table of Contents

  1. Introduction
  2. Urban-MAS
  3. Experiments
  4. Main Results
  5. Ablation Study
  6. Case Study
  7. Conclusion

Figures (1)

  • Figure 1: Urban-MAS comprises three agent layers: (Left) Predictive Factor Guidance Agents prioritize factor selection; (Middle) Reliable UrbanInfo Extraction Agents ensure reliable urban information extraction through consistency checks; (Right) Multi-UrbanInfo Inference Agents integrate multi-source urban information for robust urban prediction.