cyg

A small, dependency-free Python library to query the Cygnet multilingual wordnet database.

Cygnet is a merged multilingual wordnet covering 40+ languages and ~180,000 concepts. It combines Open English WordNet, OMW Data, TUFS, OdeNet, DanNet, and several other wordnets into a single SQLite database, indexed by the Collaborative Interlingual Index (CILI).

pip install cyg

Requires Python 3.9+.

from cyg import Cygnet

db = Cygnet() 

Note : Cygnet() automatically installs the the latest version of the database the first time you use it. Once it is done, it will always check for a new release of the database and downloads it automatically.

Counting all concepts

c = db.concepts()
print(len(c))

Filtering concepts by POS

c = db.concepts(pos="noun")

Filtering concepts by language(s)

c = db.concepts(langs="it")

Filtering concepts by form (exact match)

c = db.concepts(form="bal")

Filtering concepts by form (partial match)

c = db.concepts(contains="bal")
c = db.concepts(startswith="ball")

Filtering concepts by POS and language(s)

c = db.concepts(pos="adv", langs="it")

Filtering concepts by form and language(s)

c = db.concepts(form="bal", langs="nl")

Filtering concepts by form, POS and language(s)

c = db.concepts(form="bal", pos="noun", langs="fr")

Displaying the POS of a specific concept

c = db.concepts(form="bal")
for a in c:
    print(a.pos())

Displaying the index of a specific concept

c = db.concepts(form="bal")
for a in c:
    print(a.index())

Displaying available definitions of a concept

c = db.concepts(form="bal")
for item in c:
    f = item.definition("en")
    print(f.text())

Finding specific related concepts

c = db.concepts(form="cheek", langs="en")
for a in c:
    rel = a.hypernyms()
    print(rel)

Available relationships:

• hypernymy: a.hypernyms()
• hyponymy: a.hyponyms()
• meronymy: a.meronyms()
• holonymy: a.holonyms()

Finding all related concepts

c = db.concepts(form="leave", langs="en")
for a in c:
    all_related = a.get_related()
    print(all_related)

One or more relationships can also be specified to restrict the results.

c = db.concepts(form="book", langs="en")
for a in c:     
     g = a.get_related(['mero_member','mero_part'])
     print (g)

Retrieving senses from concepts

c = db.concepts(form="cheek", langs="en")
for a in c:
    senses = a.senses("en")
    print(a, senses)

Retrieving lexemes from concepts

c = db.concepts(form="repair", langs="en")
for a in c:
    words = a.lexemes("en")
    print(a, words)

Counting all lexemes

c = db.lexemes()
print(len(c))

Filtering lexemes by form (exact match)

c = db.lexemes(form="but")
print(len(c))

Filtering lexemes by form (partial match)

c = db.lexemes(contains="but")
d = db.lexemes(startswith="but")
print(len(c), len(d))

Filtering lexemes by language(s)

c = db.lexemes(langs=["en", "fr"])
print(len(c))

Filtering lexemes by form and language(s)

c = db.lexemes(form="but",langs=["en", "fr"])
print(c)

Finding the index of a lexeme

c = db.lexemes(form="cheek", langs="en")
for w in c:
    print (w.index())

Finding the language of a lexeme

c = db.lexemes(form="sudden")
for w in c:
    print (w.lang())

Displaying the lemma of a lexeme

c = db.lexemes(form="been")
for w in c:
    print(w.lemma())

Finding all forms of a lexeme

c = db.lexemes(form="been")
for w in c:
    print(w.all_forms())

Retrieving the concepts related to a lexeme

c = db.lexemes(form="pray")
for w in c:
    print(w.concepts())

Retrieving the senses related to a lexeme

c = db.lexemes(form="pray")
for w in c:
    print(w.senses())

Counting all senses

c = db.senses()
print(len(c))

Filtering senses by form (exact match)

c = db.senses(form="chaos")
print(c)

Filtering senses by form (partial match)

c = db.senses(contains="chaos")
d = db.senses(startswith="chaos")
print(c, d)

Filter senses by language(s)

c = db.senses(langs=["en", "fr"])
print(len(c))

Filter senses by form and language(s)

c = db.senses(form="chaos", langs=["en", "fr"])
print(c)

Displaying the index of a sense

c = db.senses(form="chaos", langs=["en", "fr"])
for b in c:
    print(b, b.index())

Displaying available examples and offsets of a sense

c = db.senses(form="plain")

for sense in c:
    for example in sense.examples():
        print(example.text(), example.sense_offsets())

Retrieving the concept related to a sense

c = db.senses(form="chaos", langs=["en"])
for b in c:
    print(b, b.concept())

Retrieving the lexeme related to a sense

c = db.senses(form="chaos")
for b in c:
    print(b, b.lexeme())

Displaying all languages included in the database

c = db.langs()

print (c)

The database is resolved in this order:

1. Explicit db_path argument to Cygnet(db_path="/path/to/cygnet.db")
2. CYG_DB environment variable
3. Platform cache directory:
   • Linux: ~/.cache/cyg/cygnet.db
   • macOS: ~/Library/Caches/cyg/cygnet.db
   • Windows: %LOCALAPPDATA%\cyg\cygnet.db

The database (~160 MB compressed) is downloaded automatically on first use from the latest GitHub release.

To disable auto-download and require an existing file:

cyg = Cygnet(download=False)  # Raises DatabaseNotFoundError if missing

Cygnet can install and manage multiple releases side by side. By default, all downloads are stored in a versioned directory layout (<cache>/cyg/<tag>/cygnet.db), so switching between releases never re-downloads the same data.

Installing a specific version:

from cyg import Cygnet

cyg = Cygnet(version="2026.05.12")

Listing available releases:

import cyg

for tag in cyg.releases():
    print(tag)  # "2026.05.12", "2026.03.15", ...

Main entry point for querying the database.

cyg = Cygnet(db_path: str | None = None, *, download: bool = True, version: str | None = None)

Filter parameters:

• form: exact match
• langs: Single language code (e.g., "en") or list (e.g., ["en", "fr"])
• pos: Part of speech ("noun", "verb", "adj", "adv", "adp", "unk", "conj", "nref")
• startswith: Prefix match on normalized form
• contains: Substring match on normalized form

All query methods return lists of model objects (never None for collections).

A language-independent concept (synset).

A pairing of a lexeme with a concept (one meaning of a word).

A word in a specific language with all its inflected forms.

A text string (definition or example) with optional sense annotations.

MIT License - see LICENSE for details.

• Online repository: https://github.com/omwn/cyg
• Cygnet database: https://github.com/omwn/cygnet
• Online interface: https://cygnet.maudslay.eu/

Maudslay, R. H., & Bond, F. (2026). Cygnet: Refactoring the Open Multilingual Wordnet. In Proceedings of the Fifteenth Language Resources and Evaluation Conference (LREC 2026) (pp. 7905–7917). European Language Resources Association (ELRA). https://doi.org/10.63317/2v96snyewr2h.