What’s Out There
← Back to swiping
Language & textFreeAcademic

DBnary: Wiktionary lexical data in RDF (all languages)

Sérasset (Université Grenoble Alpes)

Wiktionary dictionaries from 26 language editions extracted as linked data, twice a month.

An extract of Wiktionary data from many language editions modeled with the Ontolex vocabulary and published in RDF. It holds one version per Wikimedia dump, twice a month, and in 2024 supported 26 language editions. Used in computational linguistics and multilingual lexical resources.

Access

Cost
FreeFree download from Zenodo; no account needed.
License
Open licenseCC BY 4.0
Sign-up
No sign-up
How you get it
Bulk downloadTGZ (RDF)

What's in it

One row is
One lexical entry, sense, or translation in RDF triples
Covers
Global (languages)
Years
Wikimedia dumps since 2017 (semi-monthly)
Updated
Irregular
Size
26 language editions, millions of entries
lexical entrypart of speechsense definitiontranslationlanguage editionetymology

Quality

4 of 5

Actively maintained pipeline with a long track record; quality mirrors Wiktionary's unevenness.

  • Extraction errors in odd entries
  • Wiktionary editions vary widely in size

Connects to

  • Wiktionary: Source dictionary
  • Glottolog: Language identifiers
  • BabelNet: Larger multilingual semantic network

Questions it could answer

  • Which language pairs have the most translation links?
  • How large is each Wiktionary edition?

Similar data sets

  • Wikipedia Database Dumps

    Every article and every past revision of every Wikipedia, free to download.

    Language & textWikimedia FoundationFree

  • Wikidata JSON Dumps

    124 million things, from people to proteins, described as public-domain structured facts.

    Language & textWikimedia FoundationFree

  • Universal Dependencies Treebanks

    Sentences in 150+ languages, each word tagged and linked to its grammatical head the same way.

    Language & textUniversal Dependencies consortiumFree

  • FineWeb2

    Pretraining text for over 1,000 languages, with the pipeline tuned and validated language by language.

    Language & textHugging Face (FineWeb team)Free

  • OPUS Parallel Corpora

    Over 100 billion aligned sentence pairs across more than 1,000 languages.

    Language & textUniversity of Helsinki (Language Technology Research Group)Free

  • FinePDFs

    About 3 trillion tokens extracted from PDFs, a source most web corpora skip because extraction is costly.

    Language & textHugging Face (FineWeb team)Free