PlaceFM: A Training-free Geospatial Foundation Model of Places

2025-06-25

arXiv.org

AI Summary

Existing foundational models struggle to flexibly model “places”—spatially structured, semantically rich, multi-scale point-of-interest (POI) regions—hindering advances in geospatial intelligence. To address this, we propose a pretraining-free graph compression framework that constructs a nationwide heterogeneous POI graph by integrating Foursquare and OpenStreetMap data. Leveraging structure-preserving, unsupervised graph compression, our method generates multi-granularity place embeddings without reliance on large-scale pretraining. The resulting embeddings are plug-and-play, general-purpose representations of places. Evaluated on diverse downstream tasks—including urban functional zoning and regional semantic parsing—our approach demonstrates strong generalization, cross-granularity consistency, and computational scalability. It thus establishes a novel paradigm for lightweight, robust, and interpretable geospatial foundation models.

Technology Category

• Knowledge Representation and Reasoning: Geometric, Spatial, and Temporal Reasoning
• Machine Learning: Graph-based Machine Learning
• Computer Vision: Other Foundations of Computer Vision

Application Category

• Graph Algorithms and Modeling for the Web: Graph embeddings and representation learning for Web-related graphs
• Semantics and Knowledge: Scalable techniques for the creation, curation, publication, maintenance, and consumption of large, Web-based, structured, reusable, knowledge graphs and ontologies
• Web Mining and Content Analysis: Large pretrained models with web data

Abstract

Spatial structure is central to effective geospatial intelligence systems. While foundation models have shown promise, they often lack the flexibility to reason about places, which are context-rich regions spanning different spatial granularities. We propose PlaceFM, a spatial foundation model that captures place representations using a training-free graph condensation method. PlaceFM condenses a nationwide POI graph built from integrated Foursquare and OpenStreetMap data in the U.S., generating general-purpose embeddings of places. These embeddings can be seamlessly integrated into geolocation data pipelines to support a wide range of downstream tasks. Without requiring pretraining, PlaceFM offers a scalable and adaptable solution for multi-scale geospatial analysis.

Problem

• Captures place representations using training-free clustering approach
• Generates general-purpose embeddings for multi-granular geospatial analysis
• Enables scalable region-level representation without costly pre-training

Innovation

• Training-free clustering approach for place representations
• Summarizes POI graphs to generate region embeddings
• Enables scalable multi-granular geospatial analysis without pre-training