Wiktionary:Votes/2017-11/Placing Wikidata ID in sense ID of proper nouns

Voting on: Placing Wikidata ID (e.g. Q174193) in sense ID of proper nouns. For example, changing

# A ] and ] ] in ] comprising the four countries of ], ] and ] in the island of ], and ] in the island of ]. Since 1922.
# {{lb|en|historical}} The ] (1801–1922)
# {{lb|en|historical|informal}} The ] (1707-1801)

to

# {{senseid|en|Q145}}A ] and ] ] in ] comprising the four countries of ], ] and ] in the island of ], and ] in the island of ]. Since 1922.
# {{senseid|en|Q174193}}{{lb|en|historical}} The ] (1801–1922)
# {{senseid|en|Q161885}}{{lb|en|historical|informal}} The ] (1707-1801)

The template used is {{senseid}}.

Schedule:

• Vote starts: 00:00, 19 November 2017 (UTC)
• Vote ends: 23:59, 18 December 2017 (UTC)
• Vote created: Dan Polansky (talk) 09:28, 12 November 2017 (UTC)

Discussion:

• Wiktionary:Beer parlour/2017/June#How can we use Wikidata's existing data pool?
• Wiktionary:Beer parlour/2017/June#Wikidata items as senseids
• Wiktionary:Beer parlour/2017/September#Proposed first use of Wikidata: categorising planets
• Wiktionary:Beer parlour/2017/November#Vote: Placing Wikidata ID in sense ID of proper nouns
• Wiktionary talk:Votes/2017-11/Placing Wikidata ID in sense ID of proper nouns

Support I can see this being very useful for a variety of NLP tasks, plus it makes Wiktionary more machine-readable. Thomas Foster (talk) 11:14, 21 November 2017 (UTC)

Struck as ineligible to vote. —Granger (talk · contribs) 11:41, 21 November 2017 (UTC)

1. Support. My opinion is as follows. Basically, any ID is fine. The sense "Paris, France" could be "parisfrance" or "paris1" or "propernoun_francecapital" or whatever. But Wikidata already has a database with a consistent ID format, linking each place to additional information and wiki links so we might as well use the Wikidata ID. So Q90 would be the sense ID for "Paris, France" and Q830149 the sense ID for "Paris, Texas". I like that the same ID can be used for all languages, so Dutch Parijs can use "Q90" too.--Daniel Carrero (talk) 18:22, 21 November 2017 (UTC)
2. Support This, that and the other (talk) 03:31, 25 November 2017 (UTC)
3. Support -Xbony2 (talk) 02:54, 26 November 2017 (UTC)
4. Support Hopefully I won't regret this. Semantic info is definitely useful but I am also keen that wiki markup remains human-editable. Equinox ◑ 07:11, 26 November 2017 (UTC)
5. Support --Polyglot (talk) 05:56, 30 November 2017 (UTC)
6. Support As long as I don’t have to use it. I don’t buy the argument “Can look intimidating to new contributors who would have to figure out what Wikidata is to understand what these numerical identifiers are.” You look into the documentation to find out what it is. Else one already has to study rules and documentations when starting to edit any wiki without high risk of reprimand, and one is not necessarily distracted by it, {{lb}} and quotes and usage examples of all kinds everywhere already inhibit the readability. And there is syntax highlighting, though it is in beta state. Palaestrator verborum (loquier) 21:21, 9 December 2017 (UTC)

   I think it would look intimidating to me if I were a new editor. I would not want to read all sorts of documentation and "rules" like I am expected to in some cultures; I would love to start by following the model of example entries that I find. I would not probably even read the doc of {{lb}}; I would see {{lb|cs|informal}} in the markup, and say, hey, that's how they mark up Czech informal. The argument of the form "we have some elements that decrease readibility => let's indiscriminatedly add more such elements" is obviously flawed; each further decrease of readability should be based on some informal cost-benefit analysis. It's like saying, we already pollute environment => any additional pollution is fine. --Dan Polansky (talk) 13:43, 10 December 2017 (UTC)

   But how many entries have glosses that aren’t too vague to include Wikidata IDs? The entries that can have Wikidata IDs are regularly boring entries that are created and edited by regular editors. Country names, plant names, decade names. They work on soup names, or can these have Wikidata too? … Palaestrator verborum (loquier) 14:31, 10 December 2017 (UTC)

   I don't have a number, but it seems a significant portion of specific entities can have a Wikidata ID. Apart from country names that you mention, place names alone include names of cities, towns, villages, rivers, lakes, mountains, you name it; there are astronomical specific entities, etc. I don't see why new contributors should not want to work on these kinds of specific entities; 91.61.108.243, which you above picked, is hardly representative of new editors. --Dan Polansky (talk) 15:05, 10 December 2017 (UTC)

7. Support —Rua (mew) 15:21, 10 December 2017 (UTC)
8. Strong Support to foster integration between language editions and across WMF projects. —Justin (koavf)❤T☮C☺M☯ 04:06, 16 December 2017 (UTC)
9. Support for the same reason as Koavf above. Tetromino (talk) 19:41, 17 December 2017 (UTC)

1. Oppose Makes the wikitext harder to read and skim for very little benefit to Wiktionary readers and editors. Can look intimidating to new contributors who would have to figure out what Wikidata is to understand what these numerical identifier are. Let this be first tried by Wiktionaries of cultures that love numerical identifiers and ridig structures, and let's first see how well it works for them, and experience the benefits (if any) there first. --Dan Polansky (talk) 09:11, 26 November 2017 (UTC)
2. Oppose Ew. mellohi! (僕の乖離) 03:45, 27 November 2017 (UTC)
3. Oppose --WikiTiki89 17:43, 1 December 2017 (UTC)
4. Oppose I'm all for making Wiktionary more machine-readable, but not at the expense of human-readability. —Aryaman^{A (मुझसे बात करें • योगदान)} 20:58, 9 December 2017 (UTC)
5. Oppose Wyang (talk) 15:25, 10 December 2017 (UTC)
6. Oppose, unless we're going to express the labels in Solresol. (This is my way of expressing that trying to enumerate senses in a fixed manner is both inherently non-bijective and also completely incomprehensible for someone who doesn't already know the language. Because yes, this proposal is so offensive that I'm actually voting on it.) — ObſequiousNewt — Geſpꝛaͤch — Beÿtraͤge 17:18, 13 December 2017 (UTC)
7. Oppose, I don't trust Wikidata. --Robbie SWE (talk) 07:52, 15 December 2017 (UTC)
8. Oppose, due to human readability concerns and abuse of {{senseid}}. I could support this if the wikidata info could be rendered invisible, without causing further overburdening of entry downloading due to excessive Javascript. DCDuring (talk) 23:12, 16 December 2017 (UTC)

Abstain Is there no other place to put these identifiers? --Per utramque cavernam (talk) 15:09, 10 December 2017 (UTC)

Struck as ineligible to vote. —Granger (talk · contribs) 17:38, 10 December 2017 (UTC)

They have to be tied to individual senses (which are of course editable and rearrangeable by users), so I don't see where else we could safely store them. Equinox ◑ 19:03, 10 December 2017 (UTC)

1. Abstain I have a feeling that this is likely to cause problems down the road, since dictionary definitions don't seem like they'll match up perfectly with Wikidata items. I can't identify any particular reason why this would cause problems—it's just an intuitive feeling, which is why I'm abstaining rather than opposing. —Granger (talk · contribs) 17:36, 10 December 2017 (UTC)
2. Neutral It seems to me that "{{senseid|en|Q161885}}" is far more human readable and even logical than "{{lb|en|historical|informal}}". I feel it would be unsupportable to accept the local templatification and reject the senseid on that basis. That said, these suggested 'sense identification numbers' are not to a shared definition, but to the node equivalent of a Wikipedia article - a disorganized and therefore often inaccurate collation of data. E.g. the citizens of the island of Ireland were certainly not using the demonym "British" during the United Kingdom of Great Britain and Ireland. It is like using the Dewey Decimal system to create unique sense ids; extremely unuseful. Since Wikidata is already replicating Wiktionary's glosses, these should link to a proper noun gloss, and only that. - Amgine/ ^t·e 16:33, 13 December 2017 (UTC)

No consensus, 9–8–2. No action taken. —Μετάknowledge^{discuss/deeds} 08:33, 23 December 2017 (UTC)