CC0 1.0 Universal (Public Domain) Structured data

Wikidata is Wikipedia's structured-data sister project. We ingest its food subgraph — dish entities (Q-IDs), their cuisine classifications, country-of-origin links, and main-ingredient relationships.

Approximately 50,000 dish entities with their cuisine (P2012), country of origin (P495), main food ingredient (P186), and English/native-language labels. Also cuisine-to-region mappings.

The data drives our ontology-consistency check: if Wikidata records that Pad Thai's main ingredient is rice noodles, and a candidate recipe titled "Pad Thai" generates wheat pasta, we flag the mismatch. Wikidata is also a primary source for our canonical-dish discovery pipeline.

We normalize labels to lowercase for case-insensitive matching. We filter to food-related entities only (instance of dish Q746549 or subclasses). No content is modified — only selected and indexed.

Wikidata is released under the CC0 1.0 Universal Public Domain Dedication, which waives all copyright. Attribution is not legally required; we credit Wikidata anyway out of respect for the open-knowledge ecosystem. Source: wikidata.org.

CC BY-SA 4.0 Factual grounding

We query Wikipedia's REST API to confirm whether a dish has a dedicated article — a strong signal that the dish is real and documented. We do not copy article text.

Article titles, article URLs, Wikidata Q-ID sitelinks, and category-tree membership (for example, "Category:Italian cuisine" as a seed list of dish candidates). All are factual identifiers.

Dish existence grounding: if a candidate dish has a corresponding Wikipedia article, we treat that as strong evidence the dish is real. Article absence is a weaker signal because Wikipedia coverage of regional cuisines is uneven. Category trees also seed our initial dish discovery for each cuisine.

We do not reproduce any article text. We cache article-existence lookups for performance, store canonical URLs as references, and combine Wikipedia evidence with other sources to produce our own evaluations. All derivative analysis is our own.

Wikipedia article text is licensed under Creative Commons Attribution-ShareAlike 4.0. Since we reference articles rather than reproduce text, strict attribution under this license is not triggered — but we credit Wikipedia here and link to every article we reference. Source: wikipedia.org.

Open for research & education Cultural inventory

UNESCO maintains inventories of culinary traditions inscribed as Intangible Cultural Heritage of Humanity, such as the Mediterranean diet, Gastronomic meal of the French, Washoku (Japanese traditional dietary cultures), and others.

The names and descriptions of culinary traditions on the UNESCO ICH lists, used to surface long-tail traditional foods that may be underrepresented in other sources.

Seed source for canonical-dish discovery in cuisines where Wikipedia/Wikidata coverage is sparse. Cross-referenced against our other sources; UNESCO-listed foods receive a higher confidence tier in our validation pipeline.

UNESCO inventory data is made available for research, educational, and non-commercial purposes per the UNESCO Terms of Use. We acknowledge UNESCO as the source of this data. Source: ich.unesco.org.

Factual references Endangered traditional foods

The Slow Food Ark of Taste catalogues endangered traditional foods and regional culinary specialties at risk of extinction. It's an invaluable source for long-tail regional cuisine documentation.

Product names, country/region attributions, and category classifications. We use the catalogue as a factual reference to identify authentic regional specialties; we do not reproduce the descriptive text from product pages.

High-confidence source for dishes in under-documented regional cuisines. If a dish appears in both the Ark of Taste and Wikidata, it passes our canonical-source threshold automatically.

Ark of Taste content is © Slow Food. We reference products by name and link back to their Slow Food product pages. We do not reproduce descriptive content. Source: fondazioneslowfood.com.