And just to point out - even though there are no plans to accommodate the superstructures in the data model directly, it should be noted that the current data model already is flexible to have it, i.e. if the community so wishes they can create Lexemes which represent the "root" of a word like "produc-" and then explicitly link these with statements from the Lexemes for "production", "producer", etc. Or not. It could instead try to model it with statements pertaining the etymology of the words. Or not.

The Wiktionary data model is not supposed to express a specific theory of linguistics, just as the Wikidata data model is not supposed to express a specific theory of ontology. It is supposed to be flexible enough to work with whatever the community decides it wants to express, sometimes even contradictory statements, and the ability to source them to references.



On Mon, Sep 19, 2016 at 6:05 AM Daniel Kinzler <daniel.kinzler@wikimedia.de> wrote:
Am 16.09.2016 um 20:46 schrieb Thad Guidry:
> Daniel,
>
> I wasn't trying to help solve the issues - I'll be quite now :)
>
> ​I was helping to expose one of your test cases :)​

Ha, sorry for sounding harsh, and thanks for pointing me to "product"! It's a
good test case indeed.

> 'product' is a lexeme - a headword - a basic unit of meaning that has a 'set of
> forms' and those have 'a set of definitions'

In the current model, a Lexeme has forms and senses. Forms don't have senses
directly, the meanings should apply to all forms. This means lexemes have to be
split with higher granularity:

* product (English noun) would be one lexeme, with "products" being the plural
form, and "product's" the genitive, and "products'" the plural genitive. Sense
include the ones you mentioned.
* (to) produce (English verb) would be another lexeme, with forms like
"produces", "produced", "producing", etc, and senses meaning "to create", "to
show", "to make available", etc
* production (English noun) would be another lexeme, with other forms and senses.
* produce (English noun) would be another
* producer (English noun) would be another
* produced (English adjective) another
etc...

These lexemes can be linked using some kind of "derived from" statements.

> ​But a thought just occured to me...
> A. In order to model this perhaps would be to have those headwords stored in
> Wikidata.  Those headwords ideally would not actually be a Q or a P ... but what
> about instead ... L​  ?  Wrapping the graph structure itself ?  Pros / Cons ?

That's the plan, yes: Have lexemes (L...) on wikidata, which wrap the structure
of forms and senses, and has statements for the lexeme, as well as for each form
and each sense.

We don't currently plan a "super-structure" for wrapping derived/related lexemse
(product, produce, production, etc). They would just be inter-linked by statements.

> B.  or do we go with Daniel's suggestion of linking out to headwords and not
> actually storing them in Wikidata ?  Pros / Cons ?

The link I suggest is between items (Q...) and lexemes (L...), both on Wikidata.

--
Daniel Kinzler
Senior Software Developer

Wikimedia Deutschland
Gesellschaft zur Förderung Freien Wissens e.V.

_______________________________________________
Wikidata mailing list
Wikidata@lists.wikimedia.org
https://lists.wikimedia.org/mailman/listinfo/wikidata