Wikidata:WikiProject Occupations

WikiProject Occupations

This WikiProject coordinates the systematic enrichment of occupation items (occupation (P106), occupation (Q12737077)) on Wikidata using verified data from official national occupation classification systems issued by state governments.

Why this matters for Wikidata

Wikidata positions itself as a multilingual knowledge base. For occupation data, that claim does not currently hold on most languages — and the gap propagates into every Wikipedia and every downstream consumer (search engines, knowledge graphs, voice assistants, infobox templates).

The leverage point. Wikidata contains roughly 10 million items about people. Nearly all of them carry occupation (P106) (occupation) — a pointer to an occupation Q-item. When the target Q-item lacks a label in a reader's language, that reader sees an English string in place of the word they speak. One label added at the level of an occupation Q-item improves dozens-to-thousands of biographies at once. This is the highest-ROI work I have found on Wikidata.

Where biographies break visibly:

Nobel laureates — this project started from enrichment work on Nobel laureate biographies, where I discovered that the professions of laureates (physicist, economist, chemist, writer) were missing labels in many languages. Wikipedia infobox templates fall back to English strings.
Heads of state and government — every country has Q-items for "President of X", "Prime Minister of Y", "Governor of Gagauzia" (one of many positions surfaced by national classifiers). When these office-occupations lack labels in the official language(s) of the country itself, every officeholder's biography on the national-language Wikipedia is impoverished.
Anyone with a biography — scientists, doctors, artists, engineers, journalists. Their occupation (P106) points to an occupation item; if that item is empty in their language, their biography is empty in their language.

Whole occupations missing from Wikidata. Many real, mass occupations have no Q-item at all because they exist outside the international standard:

Togolese zemidjan driver (motorcycle taxi — thousands of workers)
Guatemalan piloto de bicitaxis (cycle taxi)
Honduran conductor de moto taxi forestal (forest-area motorcycle taxi)

A biographer writing about a person in any of these occupations today has nowhere to anchor occupation (P106). The reader of any future biography of these workers loses the occupation entirely.

Volunteer coverage cannot close this gap. Wikidata's volunteer model produces labels in languages with active editors. With ~10M biographies × 400+ supported languages, the human-volunteer route alone does not scale to occupations. Government-published occupation classifiers — 165 of them, from 154 countries — already contain the authoritative native-language vocabulary for this domain, under CC0 / public-domain. Bringing those existing vocabularies into Wikidata is mechanical work that the volunteer model is uniquely bad at and that a careful bot operator is uniquely good at.

Source quality matches Wikidata standards. Every label has a reference to the act of government that published it (stated in (P248), reference URL (P854), retrieved (P813)). No machine translation is used as a publication source. Where a national classifier was unavailable for a language, that gap is shown honestly on the coverage page rather than backfilled with auto-translation.

Scope

Adding labels and aliases on occupation items from government-published registries (state classifiers)
Mapping national codes to ISCO-08 (International Standard Classification of Occupations (Q1233766)) unit groups where official crosswalks exist
Every label has a reference to the source publication (national gazette, statistical office release, ministerial order)
No machine-translated content is published as a primary label — translation tools are used only for cross-lingual duplicate detection

Data source: GSCO

The Global Standard Classification of Occupations (GSCO) is the open research dataset I have built by aggregating national occupation classifiers, with per-entry provenance metadata.

165 national occupation registries from 154 countries
231,596 occupation entries
53 source languages with native-language titles from original publications
All data verbatim from official sources — CC0 license on the dataset
Live source list with per-entry provenance: gsco.io/api/sources

Research publication: Dreshmanis, M. (2026). GSCO: The Global Standard Classification of Occupations — A Deterministic Multilingual Database to Solve the N² Cross-Table Problem in International Occupation Classification. Zenodo. doi:10.5281/zenodo.19902278 · concept DOI · CC BY 4.0 · multilingual web edition

Coverage and roadmap

Coverage table — current Wikidata coverage per language vs. GSCO availability (10,383 verified occupation Q-items as of 2026-05-13)
Roadmap — phased publication plan, by country, sequenced by data readiness

Bot tasks

Task	Description	Status
Task 1	Label enrichment from ESCO multilingual data (24 EU languages)	Completed 2026-05-05. 4,701 labels added, 0 community reverts.
Task 2	Q-item creation and label enrichment from 33 national classifiers	In preparation — Wikidata:Requests for permissions/Bot/MarisDreshmanisBot 2 (forthcoming)

Methodology corrections and incident disclosure live on the project talk page.

Key properties used

instance of (P31) = occupation (Q12737077) — instance of: occupation
subclass of (P279) — subclass of
members have occupation (P3989) — ISCO-08 code (4-digit)
stated in (P248) — stated in (classifier Q-item, where it exists)
reference URL (P854) — reference URL (source publication)
retrieved (P813) — retrieved date

Participants

Maris Dreshmanis (talk • contribs • logs) — operator, GSCO dataset author
- ORCID: 0009-0003-8151-4088
- ISNI: 0000 0004 9280 9121
- Bot account: MarisDreshmanisBot (flag granted 2026-05-05)

Feedback

Project talk page
User talk:MarisDreshmanisBot — bot-specific issues
User talk:Maris Dreshmanis — methodology questions

I respond within 24 hours to any concern raised, with operations paused while the concern is being addressed.