John, sorry, I guess I was too slow - as far as I understand you have now re-read the 13-08 proposal, which has made my last Email redundant.
https://www.wikidata.org/w/index.php?title=Wikidata_talk:Wiktionary/Developm...
I hope that the model is clear now. Thanks for your engagement! Denny
On Sun, May 17, 2015 at 12:20 PM Denny Vrandečić vrandecic@gmail.com wrote:
Daniel's answer fits exactly with the proposal (which is unsurprising, because he reviewed and certainly influenced it).
To make it clear again: the proposal on
https://www.wikidata.org/wiki/Wikidata:Wiktionary/Development/Proposals/2015... is a proposal for the tasks that need to be performed. Your questions are mostly about the data model, which was discussed earlier in the following proposal:
https://www.wikidata.org/wiki/Wikidata:Wiktionary/Development/Proposals/2013...
Since I am not sure which questions remain open, I will try to address them here again, on the risk of repeating what has been said before. Unfortunately you seem to not use the terminology as defined in the second proposal linked above, which makes the discussion unnecessarily harder than it could be. If you prefer another terminology, I would be happy if you link to a one pager describing it, so that we can effectively communicate.
How do we go from a spelled form of a lexeme at Wiktionary and to an identifier
on Wikidata?
If with "spelled form of a lexeme at Wiktionary" you mean a Form as per the proposal, then the answer is: Forms have statements, and statements may point to Items, Forms, Senses, Lexemes, etc.. The exact properties to be used in these statements are up to the community.
If with "spelled form of a lexeme at Wiktionary" you mean Lexeme as per the proposal, than the answer is: Lexems have statements, and statements may point to Items, Forms, Senses, Lexemes, etc. The exact properties to be used in these statements are up to the community.
This is already stated in the second link above.
And how do we go from one Sense to another synonym Sense?
A Sense has a set of statements, and statements may point to other Senses. The exact properties used are up to the community. So a statement with the property 'synonym' stated on a Sense could point to another Sense.
Do we use statements?
Yes.
But then only the L-identifiers can be used, so we will link them at
the Lexeme level..
No. As the second link above says, Senses and Forms also have Statements. It is not only Lexemes that have Statements.
Wiktionary is organized around homonyms while Wikipedia is organized around
synonyms, especially across languages, and I think this difference creates some of the problems.
Yes, that is why Tasks 1, 2, 9 and 10 in the proposal for the task breakdown, the first link above, deal with exactly this question.
Since Gerard stated that his question was subsumed by the above list, I hope that his question is also answered?
I am afraid that I could not write a new proposal which is significantly clearer than the current, but I can keep answering questions. But all the questions you have asked seem to be explicitly answered in the two links given above. Since I know you are smart, I am wondering what is not working in the communication right now. Did you miss the first link? Because without that it is indeed hard to fully understand the second link (but the first link is already given in the second link).
So, please, keep asking questions. And everyone else too. I would like to continue improving the proposals based on your questions and suggestions.
On Sat, May 16, 2015 at 3:46 PM John Erling Blad jeblad@gmail.com wrote:
Your description is pretty far from whats in the proposal right now. The proposal is not clear at all, so I would say update it and resubmit if for a new discussion.
On Sat, May 16, 2015 at 12:21 PM, Daniel Kinzler daniel.kinzler@wikimedia.de wrote:
Am 15.05.2015 um 01:11 schrieb John Erling Blad:
How do we go from a spelled form of a lexeme at Wiktionary and to an identifier on Wikidata?
What do you mean by "go to"? And what do you mean by "identifier on
Wikidata" -
Items, Lexemes, Senses, or Forms?
Generally, Wiktionary currently combines words with the same rendering
from
different languages on a single page. So a single Wiktionary page would correspond to several Lexeme entries on Wikidata, since Lexemes on
wikidata
would be split per language.
I suppose a Lexeme-Entry could be linked back to the corresponding
pages on the
various Wiktionaries, but I don't really see the value of that, and
sitelinks
are currently not planned for Lexeme entries. It probably makes more
sense for
the Wiktionary pages to explicitly reference the Wikidata-Lexeme that corresponds to each language-section on the page.
And how do we go from one Sense to another synonym Sense? Do we use statements? But then only the L-identifiers can be used, so we will link them at the Lexeme level..
Why can only L-Identifiers be used? Senses (and Forms) are entities and
have
identifiers. They wouldn't have a wiki-page of their own, but that's
not a
problem. The intention is that it's possible for one Sense to have a
statement
referring directly to another Sense (of the same or a different Lexeme).
Wiktionary is organized around homonyms while Wikipedia is organized around synonyms, especially across languages, and I think this difference creates some of the problems.
The Lexeme-Part of Wikidata (L-ids) would be separate from the
Concept-part of
Wikidata (Q-ids). The Lexeme part is organized around homonyms (more
precisely,
homographs in a single language). Each Lexeme can have several "Senses"
modeled
as "sub-entities", meaning that each Sense has its own set of
Statements. Each
Sense can be linked to Senses of other Lexemes (explicit synonyms or translations) and to Q-id concepts (implicit synonyms or translations)
using
Statements.
-- Daniel Kinzler Senior Software Developer
Wikimedia Deutschland Gesellschaft zur Förderung Freien Wissens e.V.
Wikidata-l mailing list Wikidata-l@lists.wikimedia.org https://lists.wikimedia.org/mailman/listinfo/wikidata-l
Wikidata-l mailing list Wikidata-l@lists.wikimedia.org https://lists.wikimedia.org/mailman/listinfo/wikidata-l