Wikidata:Property proposal/gf-wordnet-lexeme
gf-wordnet-lexeme edit
Originally proposed at Wikidata:Property proposal/Lexemes
| Description | MISSING |
|---|---|
| Data type | External identifier |
| Example 1 | apple (Q89) → https://cloud.grammaticalframework.org/wordnet#apple_1_N |
| Example 2 | points of the compass (Q11114344) → https://cloud.grammaticalframework.org/wordnet#southwest_1_N |
| Example 3 | household (Q259059) → https://cloud.grammaticalframework.org/wordnet#household_N |
| Source | https://cloud.grammaticalframework.org/wordnet |
| Number of IDs in source | about 30 000, more in the future |
Motivation edit
GF WordNet is a variant of WordNet which brings together lexicons for several languages. It is also compatible with the grammar libraries in GF which makes it possible to use the lexicon for parsing and natural language generation.
Unlike the traditional WordNet, a synset is a set of abstract word identifiers. Each identifier is then mapped to a word in a concrete language. The abstract words make it possible to preserve the translation equivalents, which is lost if only synsets are linked across languages.
Wikidata already links entities to WordNet. In fact half of the links are exported from GF WordNet. Linking to synsets, however, is not always enough to identify lexical items unambiguously. For example household_N is best linked to Q259059 but if the entire synset is linked then it also contains family_1_N, home_8_N and house_10_N.
Since we now experiment with using using GF for natural language generation, it is useful if we preserve more explicit links. A prototype for the data already exists. – The preceding unsigned comment was added by Kr.angelov (talk • contribs) at 10:44, 8 February 2023 (UTC).
Not done - not created per lack of support. Regards, ZI Jony (Talk) 12:34, 24 January 2024 (UTC)
