AddressTable
收藏官方服务：
资源简介：
A large-scale real-world address dataset (extracted from WDC) with validated ground truth, containing 9.3M addresses spanning 128 countries. The dataset includes real-world errors such as missing values, abbreviations, language variations, and typos, along with their validated ground truth. DATASET CONTENTS: - Full-unnamed version: 9.3M records (no entity names) - Full-named version: 4.3M records with entity names - Country slices: US (2.3M), Germany (375K), Russia (194K) - POI slices: 22 variants (libraries, shops, hotels, etc.) - Datasets used in cleaning systems benchmarking
提供机构：
Zenodo创建时间：
2026-06-26
