Boundary Prompting: Elastic Urban Region Representation via Graph-based Spatial Tokenization

2025-03-11

AI Summary

Traditional urban area modeling relies on static administrative boundaries, failing to capture spatiotemporal dynamics and functional heterogeneity. To address this, we propose an elastic region representation framework based on graph-structured spatial tokenization. We introduce two novel constructs: a “spatial token dictionary” and a “region token set”, unifying urban entities as graph nodes. Our method incorporates graph neural networks, multi-channel embedding aggregation, and spatial tokenization, and features a fast online token-set extraction strategy enabling flexible, boundary-prompted definition of multi-scale, task-adaptive regions. Experiments across diverse urban analytics tasks demonstrate that our framework significantly outperforms grid- and administrative-boundary-based baselines. It achieves superior generalizability, high computational efficiency, and precise modeling of both spatiotemporal dynamics and functional heterogeneity—offering a scalable, adaptive foundation for intelligent urban analysis.

Technology Category

• Knowledge Representation and Reasoning: Geometric, Spatial, and Temporal Reasoning
• Machine Learning: Graph-based Machine Learning
• Planning, Routing, and Scheduling: Optimization of Spatio-temporal Systems

Application Category

• Graph Algorithms and Modeling for the Web: Graph embeddings and representation learning for Web-related graphs
• Search and Retrieval-Augmented AI: Web query analysis, representation and understanding
• Semantics and Knowledge: Methods to enhance, augment, integrate or synergize semantic models such as knowledge graphs and LLMs

Abstract

Urban region representation is essential for various applications such as urban planning, resource allocation, and policy development. Traditional methods rely on fixed, predefined region boundaries, which fail to capture the dynamic and complex nature of real-world urban areas. In this paper, we propose the Boundary Prompting Urban Region Representation Framework (BPURF), a novel approach that allows for elastic urban region definitions. BPURF comprises two key components: (1) A spatial token dictionary, where urban entities are treated as tokens and integrated into a unified token graph, and (2) a region token set representation model which utilize token aggregation and a multi-channel model to embed token sets corresponding to region boundaries. Additionally, we propose fast token set extraction strategy to enable online token set extraction during training and prompting. This framework enables the definition of urban regions through boundary prompting, supporting varying region boundaries and adapting to different tasks. Extensive experiments demonstrate the effectiveness of BPURF in capturing the complex characteristics of urban regions.

Problem

• Dynamic urban region representation for urban planning.
• Elastic region boundaries using graph-based tokenization.
• Fast token extraction for adaptive boundary prompting.

Innovation

• Elastic urban region representation via boundary prompting
• Graph-based spatial tokenization for urban entities
• Multi-channel model for embedding region token sets