Logo
ScienceStack
Table of Contents
Fetching ...
Sign In

Spatial-RAG: Spatial Retrieval Augmented Generation for Real-World Geospatial Reasoning Questions

Dazhou Yu, Riyang Bao, Ruiyu Ning, Jinghong Peng, Gengchen Mai, Liang Zhao

TL;DR

Spatial-RAG extends retrieval-augmented generation to real-world geospatial question answering by unifying structured spatial databases with LLM reasoning. It introduces a sparse-dense hybrid retriever and a multi-objective, Pareto-frontier approach to balance spatial constraints and semantic intent, with LLM-driven dynamic trade-offs for final generation. Empirical results across TourismQA and MapQA datasets show significant gains over strong baselines, demonstrating improved accuracy, precision, and ranking while highlighting the importance of each architectural component through ablations. The framework offers a practical path toward integrating geospatial grounding into natural language QA for applications in routing, recommendations, and urban planning.

Abstract

Answering real-world geospatial questions--such as finding restaurants along a travel route or amenities near a landmark--requires reasoning over both geographic relationships and semantic user intent. However, existing large language models (LLMs) lack spatial computing capabilities and access to up-to-date, ubiquitous real-world geospatial data, while traditional geospatial systems fall short in interpreting natural language. To bridge this gap, we introduce Spatial-RAG, a Retrieval-Augmented Generation (RAG) framework designed for geospatial question answering. Spatial-RAG integrates structured spatial databases with LLMs via a hybrid spatial retriever that combines sparse spatial filtering and dense semantic matching. It formulates the answering process as a multi-objective optimization over spatial and semantic relevance, identifying Pareto-optimal candidates and dynamically selecting the best response based on user intent. Experiments across multiple tourism and map-based QA datasets show that Spatial-RAG significantly improves accuracy, precision, and ranking performance over strong baselines.

Spatial-RAG: Spatial Retrieval Augmented Generation for Real-World Geospatial Reasoning Questions

TL;DR

Spatial-RAG extends retrieval-augmented generation to real-world geospatial question answering by unifying structured spatial databases with LLM reasoning. It introduces a sparse-dense hybrid retriever and a multi-objective, Pareto-frontier approach to balance spatial constraints and semantic intent, with LLM-driven dynamic trade-offs for final generation. Empirical results across TourismQA and MapQA datasets show significant gains over strong baselines, demonstrating improved accuracy, precision, and ranking while highlighting the importance of each architectural component through ablations. The framework offers a practical path toward integrating geospatial grounding into natural language QA for applications in routing, recommendations, and urban planning.

Abstract

Answering real-world geospatial questions--such as finding restaurants along a travel route or amenities near a landmark--requires reasoning over both geographic relationships and semantic user intent. However, existing large language models (LLMs) lack spatial computing capabilities and access to up-to-date, ubiquitous real-world geospatial data, while traditional geospatial systems fall short in interpreting natural language. To bridge this gap, we introduce Spatial-RAG, a Retrieval-Augmented Generation (RAG) framework designed for geospatial question answering. Spatial-RAG integrates structured spatial databases with LLMs via a hybrid spatial retriever that combines sparse spatial filtering and dense semantic matching. It formulates the answering process as a multi-objective optimization over spatial and semantic relevance, identifying Pareto-optimal candidates and dynamically selecting the best response based on user intent. Experiments across multiple tourism and map-based QA datasets show that Spatial-RAG significantly improves accuracy, precision, and ranking performance over strong baselines.

Paper Structure

This paper contains 45 sections, 16 equations, 7 figures, 4 tables.

Table of Contents

  1. Introduction
  2. Related work
  3. Retrieval augmented generation
  4. Geospatial question and answering
  5. Problem formulation
  6. Spatial-RAG for geospatial reasoning questions
  7. Overview
  8. Sparse spatial retrieval
  9. Geometry recognition
  10. Query function recognition and parameter estimation
  11. Hybrid spatial relevance scoring
  12. Sparse spatial relevance scoring
  13. Dense spatial relevance scoring
  14. Extracting spatial requirements and ranking via cosine similarity
  15. Hybrid spatial scoring as a generalized model
  16. ...and 30 more sections

Figures (7)

  • Figure 1: A real-world spatial reasoning question with nearby spatial objects. Areas that satisfy the spatial constraint are highlighted in purple.
  • Figure 2: Illustration of the proposed Spatial-RAG framework.
  • Figure 3: Three POI queries with different reference objects.
  • Figure 4: (a) Case 1; (b) Case 2
  • Figure 5: (a) Case 3; (b) Case 4
  • ...and 2 more figures