This work unscrambles the characteristics of relative concepts, designs a graph structure and identifies modelling strategies. It proposes a schema on how to perform address matching and toponym disambiguation using an address graph.
- Homepage
- Demos
- Demo 1
- Demo 2
- Features
- Usage
- Data
- Code
- License
Please refer to the data URL.
Figshare:
- Demo 1: Address Graph in Neo4j
- Demo 2: The center node representing an address component
- 1. Different relations connection
- 2. Supporting simple and fuzzy address matching
- 3. Supporting toponym disambiguation
- Administrative division data
(corresponding section: Section 4)
The data are a sample of Chinese administrative divisions, which are used in the toponym disambiguation task. Limited by the national policy, only <Nanjing City, JiangSu Province> is provided.
In the China-Nanjing.xlsx file, each column represents a data record referring to a specific administrative division. The name field is used to create has_component relations in the address graph.
- Graph data
(corresponding section: Section 3)
The Edge and Node directories store the basic node and edge files that can be imported into the graph database through a "Load csv" command. The name of the file is the types of nodes and edges. Generally, we first import node files and then edge files.
Address.csv is a complete preparing file to generate node and edge files in the project. In the file, address segmentation has been provided according to our semiautomatic method. We hope that the offered address segmentation can support more applications.
- Matching data
(corresponding section: Section 4)
The Address-Different Type file is used in the address matching task. It provides the queried address with different changed formats. The data can be reused to test fuzzy matching.
The Address-Different Length file is used in the address matching task. It provides the queried address with different string lengths.
- PythonPro_AMatching
(corresponding section: Section 4)
The project code is a prototype demo of address matching based on an address graph including for the original format and changed format. The main.py is the entrance of all projects. Simply copy the directory into a new project PyCharm, prepare the required pylib, then test the different addresses in the Matching data.
Copyright © 2022 Address Graph Builder. This project is MIT licensed.