What question did this study set out to answer?

To develop and evaluate a multi-stage framework that separates database querying from spatiotemporal computation to improve natural language-driven geospatial analysis.

August 19, 2026ISPRS International Journal of Geo-InformationOpen Access

CGD-QCSF: A Code Generation-Driven Query–Computation Separation Framework for Natural Language Geospatial Analysis

View Full Paper

Ask AI

Bookmark

Authors

ZLZHIYUAN LE HLHao Li YMYuanxun Mei

Discussion

Member takes

Overview

Computational evaluation demonstrates high accuracy across 200 geospatial tasks, indicating that query-computation separation improves natural language spatial analysis.

Key Points

To develop and evaluate a multi-stage framework that separates database querying from spatiotemporal computation to improve natural language-driven geospatial analysis.
Designed the CGD-QCSF architecture featuring a structured planner, an execution state manager, a SQL Code Generation Agent for database queries, and a Spatiotemporal Computation Agent operating in a Python sandbox.
Benchmarked the framework against 200 geospatial tasks across easy, medium, and hard difficulty levels using Qwen3.7-Plus as the foundation model.
CGD-QCSF achieved a Strict Structured Accuracy of 90.5% across the 200 benchmark tasks.
Ablation testing demonstrated that removing the Planner Agent reduced Strict Structured Accuracy to 84.5%, while removing the Spatiotemporal Computation Agent reduced it to 70.5%.

Cite This Study

LE et al. (2026) studied this question.

synapsesocial.com/papers/6a858aaf03308d306e2d7efd https://doi.org/10.3390/ijgi15080370

View Full Paper

Ask AI

Bookmark

August 19, 2026ISPRS International Journal of Geo-InformationOpen Access

CGD-QCSF: A Code Generation-Driven Query–Computation Separation Framework for Natural Language Geospatial Analysis

View Full Paper

Ask AI

Bookmark

Authors

ZLZHIYUAN LE HLHao Li YMYuanxun Mei

Discussion

Member takes

Overview

Computational evaluation demonstrates high accuracy across 200 geospatial tasks, indicating that query-computation separation improves natural language spatial analysis.

Key Points

To develop and evaluate a multi-stage framework that separates database querying from spatiotemporal computation to improve natural language-driven geospatial analysis.
Designed the CGD-QCSF architecture featuring a structured planner, an execution state manager, a SQL Code Generation Agent for database queries, and a Spatiotemporal Computation Agent operating in a Python sandbox.
Benchmarked the framework against 200 geospatial tasks across easy, medium, and hard difficulty levels using Qwen3.7-Plus as the foundation model.
CGD-QCSF achieved a Strict Structured Accuracy of 90.5% across the 200 benchmark tasks.
Ablation testing demonstrated that removing the Planner Agent reduced Strict Structured Accuracy to 84.5%, while removing the Spatiotemporal Computation Agent reduced it to 70.5%.

Cite This Study

LE et al. (2026) studied this question.

synapsesocial.com/papers/6a858aaf03308d306e2d7efd https://doi.org/10.3390/ijgi15080370

Also Consider

Synapse has enriched 5 closely related papers on similar clinical questions. Consider them for comparative context:

View Full Paper

Ask AI

Bookmark