Published June 26, 2026
| Version v1.0-2026-06-26
Dataset
Open
AddressTable
Authors/Creators
-
1.
Technische Universität Berlin
-
2.
Berlin Institute for the Foundations of Learning and Data
-
3.
Hasso Plattner Institute
Description
AddressTable: A Large-Scale Real-World Address Dataset for Data Cleaning Benchmarking
This dataset accompanies the paper:
Clean Me If You Can: A Large Collection of Real-World Addresses for Data Cleaning Benchmarking
Fatemeh Ahmadi, Tobias Bernhard, Mohamed Abdelmaksoud, Luca Zecchini, Tilmann Rabl, Ziawasch Abedjan (GitHub Repo)
Overview
AddressTable is a large-scale, real-world address dataset with validated ground truth for benchmarking data cleaning systems. It contains postal addresses of public entities (businesses, hotels, libraries, hospitals, etc.) extracted from the Web Data Commons (WDC) corpus, paired with cleaned ground-truth records derived from geocoding services and refined through postprocessing
This dataset captures naturally occurring data quality issues: missing values, abbreviations, language inconsistencies, encoding errors, typos, and more across 128 countries.
Dataset Contents
The archive `AddressTable.zip` (7.3 GB) contains the following datasets:
Full Versions
| Dataset | # Records | Description |
| full-named |
4,338,109
|
Addresses with entity names (e.g., hotel or library name)
|
| full-unnamed |
9,317,886
|
Addresses without entity names
|
Each dataset is provided in both a dirty version and a ground-truth version.
Country Slices (from `full-named`)
| Slice | # Records |
| us-slice | 2,267,895 |
| de-slice |
374,533
|
| ru-slice |
194,274
|
POI Slices
22 point-of-interest slices reflecting Schema.org entity types (e.g., `library`, `shopping-center`, `hotel`, `hospital`). See the paper and GitHub repository for the full list.
Benchmark Samples
Samples used in the cleaning experiments reported in the paper, at varying error rates (5%, 10%, 20%, and original), for `full-named`, `de-slice`, `ru-slice`, `library`, and `shopping-center`.
Schema
Each record contains the following attributes:
| Attribute | Description |
| _id |
Unique record identifier
|
| name |
Entity name (full-named version only)
|
| house_number |
House/building number
|
| road |
Street name
|
| postcode |
Postal code
|
| locality |
City, town, or village
|
| region |
State, province, or administrative region
|
| country | Country name |
| country_code |
ISO 3166-1 alpha-2 country code
|
| graph_iri |
Source page URL (provenance)
|
Error Statistics
Approximately 49% of cells in the full-named dataset contain errors. Errors are categorized into 12 types:
| Category | Description | Example |
| Missing |
Empty value
|
`""` ↔ `"Berlin"`
|
| Parsing |
Encoding artifacts
|
`"M%C3%BCnchen"` ↔ `"München"`
|
| Capitalization |
Case differences
|
`"ATLANTA"` ↔ `"Atlanta"`
|
| Special Charatcters |
Punctuation differences
|
`"U.S.A."` ↔ `"USA"`
|
| Name Reformat |
Token reordering
|
`"Howson Branch, Austin Public Library"` ↔ `"Austin Public Library – Howson Branch"`
|
| Language |
Translation
|
`"Austria"` ↔ `"Österreich"`
|
| Abbreviation |
Shortened forms
|
`"Ave"` ↔ `"Avenue"`, `"Bachmannstr."` ↔ `"Bachmannstraße"`
|
| Token Subset |
Partial names
|
`"New York"` ↔ `"New York City"`
|
| Typo |
Small edit-distance differences
|
`"Theatre"` ↔ `"Theater"`
|
| Word Substitution |
Single-token factual change
|
`"VA Clinic"` ↔ `"Outpatient Clinic"`
|
|
Miscellaneous
|
All other differences - Semantic errors
|
`Library` ↔ `Hotel` |
Files
AddressTable.zip
Additional details
Dates
- Available
-
2026-06-26
Software
- Repository URL
- https://github.com/D2IP-TUB/Clean-Me-If-You-Can
- Development Status
- Active