Robust multi-source geographic entities matching by maximizing geometric and semantic similarity

Area of Science:

• Geographic Information Science
• Spatial Data Analysis
• Computer Science

Background:

• Geographic entity matching is crucial for integrating multi-source spatial data.
• Existing methods struggle with heterogeneous data and complex patterns.
• Limitations in generalization and accuracy hinder multi-source data fusion.

Purpose of the Study:

• To propose a robust method for multi-source geographic entity matching.
• To improve accuracy and generalization in matching heterogeneous spatial data.
• To enhance spatial data fusion and information sharing capabilities.

Main Methods:

• Entity segmentation based on shape features to extract matching primitives.
• Grouping feature segments into patterns (3 major, 14 subcategories).
• Matching patterns using spatial similarity metrics and refining with semantic similarity.

Main Results:

• The method effectively matches both area and line geographic entities.
• Demonstrated strong generalization and application capability across diverse datasets.
• Achieved high matching accuracy with precision and recall nearing 90%.

Conclusions:

• The proposed method offers a significant improvement over existing approaches.
• It effectively addresses challenges in matching heterogeneous multi-source geographic data.
• The approach enhances the reliability of spatial data fusion and analysis.