license: cc0-1.0
language:
- en
- fr
- de
- cs
- ja
- nl
- pl
- 'no'
- fi
- sv
- multilingual
tags:
- wikidata
- companies
- kyb
- business-registry
- knowledge-graph
- entity-linking
size_categories:
- 100K<n<1M
pretty_name: Wikidata × OpenRegistry company alignment
Wikidata × OpenRegistry company alignment
A precomputed map between Wikidata items for companies and their canonical OpenRegistry URLs.
Cite as: Sophymarine (2026). Wikidata × OpenRegistry company alignment (Version 1.0.0) [Data set]. Zenodo. https://doi.org/10.5281/zenodo.20363013
For every Wikidata item that has at least one supported national company-registry identifier statement, this dataset provides a single row giving the QID, the registry identifier, and the corresponding direct URL on OpenRegistry — the multi-registry gateway operated by Sophymarine, licensed by UK Companies House, Ireland CRO, Norway Brønnøysundregistrene, Switzerland Zefix, Corporations Canada, Netherlands KvK, Australia ABR, France Sirene, Finland PRH, Czechia ARES, Poland KRS, Belgium KBO, Isle of Man Companies Registry and others.
The alignment was built by querying the Wikidata Query Service on 2026-05-24 for each of 17 national company-identifier properties and combining the results.
Why this dataset exists
Wikidata contains structured, community-curated information about hundreds of thousands of companies, but linking from a Wikidata QID to a live registry record requires reconstructing one URL per jurisdiction (each national registry has its own URL scheme, some with extra session/POST/redirect requirements). This dataset eliminates that reconstruction step: one column gives you a working URL that returns the company record in your choice of HTML, Markdown, JSON-LD, or Turtle.
Useful for:
- Entity-linking: resolve a Wikidata company QID to a live registry record in one lookup
- Wikipedia / Wikidata reference enrichment
- KYB and supply-chain due-diligence pipelines
- Knowledge-graph backfills for AI agents (the URL also serves an MCP-friendly variant)
- Coverage analysis: which jurisdictions have how many community-validated company items
Schema
| Column | Type | Description |
|---|---|---|
wikidata_qid |
string | Wikidata QID, e.g. Q130879 |
country_code |
string | ISO 3166-1 alpha-2, lowercase |
registry_property |
string | Wikidata property used as the source, e.g. P2622 |
registry_id |
string | The registry's native identifier, verbatim |
openregistry_url |
string | Canonical OpenRegistry company URL |
wikidata_label_en |
string | English label (may be empty) |
wikidata_label_local |
string | A non-English label (may be empty) |
Coverage
| Country | Registry | Property | Rows |
|---|---|---|---|
| France | SIREN | P1616 | 462,518 |
| Japan | Corporate Number | P3225 | 86,300 |
| Czechia | IČO (ARES) | P4156 | 76,332 |
| Norway | Brønnøysundregistrene | P2333 | 10,075 |
| United Kingdom | Companies House | P2622 | 7,581 |
| Netherlands | KvK | P3220 | 4,710 |
| Switzerland | UID (Zefix) | P4829 | 3,245 |
| Australia | ABN | P3548 | 2,793 |
| Belgium | KBO | P3376 | 2,438 |
| Poland | KRS | P12228 | 1,012 |
| Canada | Business Number | P8860 | 649 |
| Ireland | CRO | P9679 | 290 |
| New Zealand | NZBN | P5687 | 139 |
| Malaysia | SSM | P6843 | 122 |
| Finland | PRH | P6775 | 64 |
| South Korea | Corporate Number | P6859 | 24 |
| Hong Kong | CR | P13942 | 5 |
| Total | 658,297 |
Examples
wikidata_qid,country_code,registry_property,registry_id,openregistry_url,wikidata_label_en,wikidata_label_local
Q130879,gb,P2622,03477402,https://openregistry.sophymarine.com/company/gb/03477402,Reuters,
Q170614,ie,P9679,246488,https://openregistry.sophymarine.com/company/ie/246488,Ryanair,
Q101341,no,P2333,945578564,https://openregistry.sophymarine.com/company/no/945578564,Lillehammer,
Q1632529,cz,P4156,00261874,https://openregistry.sophymarine.com/company/cz/00261874,,Hrušovany
The openregistry_url is a stable canonical URL. Appending .md, .jsonld, or .ttl to the URL returns the record in those formats:
- HTML:
https://openregistry.sophymarine.com/company/gb/03477402 - Markdown:
https://openregistry.sophymarine.com/company/gb/03477402.md - JSON-LD:
https://openregistry.sophymarine.com/company/gb/03477402.jsonld - Turtle:
https://openregistry.sophymarine.com/company/gb/03477402.ttl
What this dataset is NOT
- Not a complete list of companies. Wikidata only contains entities that the community has judged notable; ordinary small companies are excluded by design. UK Companies House alone has 5M+ active records; only 7,581 appear here.
- Not OpenRegistry's data. The mapping is built entirely from Wikidata's public dump and OpenRegistry's URL scheme. OpenRegistry's underlying registry data is not included.
- Not derived from OpenCorporates. Items with only a P1320 (OpenCorporates) statement and no native national-registry property are excluded from this version. Future versions may add a P1320 fallback.
- Not a real-time feed. The snapshot reflects Wikidata state on 2026-05-24. Refresh by re-running the SPARQL queries against Wikidata Query Service.
Licence
The dataset is published under CC0 1.0 Universal (public domain dedication), matching Wikidata's own licence on the underlying QIDs and registry identifiers.
OpenRegistry URLs are public addresses; reusing them carries no additional restriction. The data returned at those URLs is subject to each national registry's own terms (see each registry's licence; most use Open Government Licence variants).
Citation
@dataset{wikidata_openregistry_alignment_2026,
title = {Wikidata × OpenRegistry company alignment},
author = {Sophymarine},
year = {2026},
month = {may},
publisher = {Zenodo},
version = {1.0.0},
doi = {10.5281/zenodo.20363013},
url = {https://doi.org/10.5281/zenodo.20363013}
}
Mirrors:
- HuggingFace Datasets: https://huggingface.co/datasets/Sophymarine/wikidata-alignment
- Zenodo (with DOI): https://doi.org/10.5281/zenodo.20363013
Source
OpenRegistry — multi-registry gateway: https://openregistry.sophymarine.com Wikidata — knowledge graph: https://www.wikidata.org
Refresh script
Each per-jurisdiction TSV in this dataset was produced by a SPARQL query of the form:
SELECT ?item ?id (SAMPLE(?lEn) AS ?labelEn) (SAMPLE(?lLoc) AS ?labelLocal) WHERE {
?item wdt:P2622 ?id .
OPTIONAL { ?item rdfs:label ?lEn . FILTER(LANG(?lEn)='en') }
OPTIONAL { ?item rdfs:label ?lLoc . FILTER(LANG(?lLoc)!='en') }
} GROUP BY ?item ?id
(replace P2622 with the property for the target country). France's 462k rows required chunking by SIREN first digit to stay under the Wikidata Query Service 60-second timeout.