Bridging natural language and GIS : a multi-agent framework for LLM-driven autonomous geospatial analysis
Mansourian, Ali LU
and Oucheikh, Rachid
LU
(2026)
In International Journal of Digital Earth
19(1).
- Abstract
Existing LLM-based approaches remain limited by simplistic task execution, restricted tool integration, and a lack of contextual reasoning when interacting with professional GIS software. This study investigates the efficacy of a multi-agent architecture designed to enhance geospatial task execution accuracy through collaboration, reasoning and tool-use. The architecture integrates Chain of Thought (CoT) reasoning and Retrieval-Augmented Generation (RAG) and employs specialized agents that collaboratively translate high-level natural language queries into structured, executable workflows using QGIS processing algorithms as tools. Through a structured fine-tuning approach, we evaluated how the multi-agent setup influences spatial task... (More)
Existing LLM-based approaches remain limited by simplistic task execution, restricted tool integration, and a lack of contextual reasoning when interacting with professional GIS software. This study investigates the efficacy of a multi-agent architecture designed to enhance geospatial task execution accuracy through collaboration, reasoning and tool-use. The architecture integrates Chain of Thought (CoT) reasoning and Retrieval-Augmented Generation (RAG) and employs specialized agents that collaboratively translate high-level natural language queries into structured, executable workflows using QGIS processing algorithms as tools. Through a structured fine-tuning approach, we evaluated how the multi-agent setup influences spatial task comprehension, geoprocessing tool selection, and code generation. The results demonstrate that the developed framework significantly outperforms baseline single-agent and non-fine-tuned systems. For tasks involving one or two GIS tools, the system achieved up to 100% execution success and 87.5% semantic correctness. However, its effectiveness decreases with more complex, multi-step workflows. Notably, iterative self-refinement and self-debugging led to moderate gains in execution success and semantic correctness. The results indicate that multi-agent frameworks are a promising approach, even though improvements in reasoning depth and tool alignment are still needed. This study represents a foundational step toward more reliable, modular, and adaptable LLM-based systems for geospatial analysis automation.
(Less)
- author
- Mansourian, Ali
LU
and Oucheikh, Rachid
LU
- organization
-
- Department of Earth and Environmental Sciences (MGeo)
- LU Profile Area: Nature-based future solutions
- Centre for Advanced Middle Eastern Studies (CMES)
- MECW: The Middle East in the Contemporary World
- BECC: Biodiversity and Ecosystem services in a Changing Climate
- Centre for Geographical Information Systems (GIS Centre)
- Dept of Physical Geography and Ecosystem Science
- eSSENCE: The e-Science Collaboration
- publishing date
- 2026
- type
- Contribution to journal
- publication status
- published
- subject
- keywords
- AI Agent, autonomous geospatial analysis, GeoAI, GIS, Large Languages Models (LLMs)
- in
- International Journal of Digital Earth
- volume
- 19
- issue
- 1
- article number
- 2633849
- publisher
- Taylor & Francis
- external identifiers
-
- scopus:105030928762
- ISSN
- 1753-8947
- DOI
- 10.1080/17538947.2026.2633849
- language
- English
- LU publication?
- yes
- additional info
- Publisher Copyright: © 2026 The Author(s). Published by Informa UK Limited, trading as Taylor & Francis Group.
- id
- 3108a8ae-165b-4ec7-93c3-15355c4771d8
- date added to LUP
- 2026-04-21 14:33:28
- date last changed
- 2026-06-02 17:20:50
@article{3108a8ae-165b-4ec7-93c3-15355c4771d8,
abstract = {{<p>Existing LLM-based approaches remain limited by simplistic task execution, restricted tool integration, and a lack of contextual reasoning when interacting with professional GIS software. This study investigates the efficacy of a multi-agent architecture designed to enhance geospatial task execution accuracy through collaboration, reasoning and tool-use. The architecture integrates Chain of Thought (CoT) reasoning and Retrieval-Augmented Generation (RAG) and employs specialized agents that collaboratively translate high-level natural language queries into structured, executable workflows using QGIS processing algorithms as tools. Through a structured fine-tuning approach, we evaluated how the multi-agent setup influences spatial task comprehension, geoprocessing tool selection, and code generation. The results demonstrate that the developed framework significantly outperforms baseline single-agent and non-fine-tuned systems. For tasks involving one or two GIS tools, the system achieved up to 100% execution success and 87.5% semantic correctness. However, its effectiveness decreases with more complex, multi-step workflows. Notably, iterative self-refinement and self-debugging led to moderate gains in execution success and semantic correctness. The results indicate that multi-agent frameworks are a promising approach, even though improvements in reasoning depth and tool alignment are still needed. This study represents a foundational step toward more reliable, modular, and adaptable LLM-based systems for geospatial analysis automation.</p>}},
author = {{Mansourian, Ali and Oucheikh, Rachid}},
issn = {{1753-8947}},
keywords = {{AI Agent; autonomous geospatial analysis; GeoAI; GIS; Large Languages Models (LLMs)}},
language = {{eng}},
number = {{1}},
publisher = {{Taylor & Francis}},
series = {{International Journal of Digital Earth}},
title = {{Bridging natural language and GIS : a multi-agent framework for LLM-driven autonomous geospatial analysis}},
url = {{http://dx.doi.org/10.1080/17538947.2026.2633849}},
doi = {{10.1080/17538947.2026.2633849}},
volume = {{19}},
year = {{2026}},
}