Wikidata:Property proposal/same as (permanently duplicated item)
permanent duplicated item edit
| Data type | Item |
|---|---|
| Allowed values | instances of Wikimedia permanent duplicate item (Q21286738) or item linked with this property from an item with instances of Wikimedia permanent duplicate item (Q21286738) |
| Example | Universe (Q1) → Universe (Q22924128), Universe (Q22924128) → Universe (Q1) |
- Motivation
Currently, items are using said to be the same as (P460) for linking to items linked to duplicate articles with different scripts. This should be split into a separate property to avoid confusion. --Yair rand (talk) 13:39, 29 March 2016 (UTC)
- Discussion
Support shouldn't this be "duplicate of item", "duplicate of" or something like that? - Brya (talk) 11:21, 30 March 2016 (UTC)
- @Brya: Not "duplicate of", I think. This property is "meta", referring to Wikidata items themselves, so it needs to be distinguished. (Duplicate items are only necessary in the first place because of Wikibase limitations.) "duplicate of item" means the same thing as "duplicate item", since it's a symmetrical relation, so I don't think it makes a difference which is used. --Yair rand (talk) 21:27, 30 March 2016 (UTC)
- Very often (always?) the relationship is not symmetrical: on the one hand there is the full item (with all the trimmings), while on the other hand there is the duplicate which holds just one sitelink (to a page that should be cleared/merged/redirected as soon as possible). - Brya (talk) 04:49, 31 March 2016 (UTC)
- From the proposal, this doesn't seem to apply to items that should be merged/redirected.
--- Jura 06:15, 1 April 2016 (UTC)- I see. OK. - Brya (talk) 10:38, 1 April 2016 (UTC)
- From the proposal, this doesn't seem to apply to items that should be merged/redirected.
- Very often (always?) the relationship is not symmetrical: on the one hand there is the full item (with all the trimmings), while on the other hand there is the duplicate which holds just one sitelink (to a page that should be cleared/merged/redirected as soon as possible). - Brya (talk) 04:49, 31 March 2016 (UTC)
- @Brya: Not "duplicate of", I think. This property is "meta", referring to Wikidata items themselves, so it needs to be distinguished. (Duplicate items are only necessary in the first place because of Wikibase limitations.) "duplicate of item" means the same thing as "duplicate item", since it's a symmetrical relation, so I don't think it makes a difference which is used. --Yair rand (talk) 21:27, 30 March 2016 (UTC)
Comment Could you add "permanent" to the label? Otherwise people use it for items that are just waiting to be merged.
--- Jura 04:08, 31 March 2016 (UTC)- @Jura1: Good idea. Done. --Yair rand (talk) 19:29, 3 April 2016 (UTC)
- Comment There is another problem. For some wikis, it's easy to figure out that there are two items about the same topic in different scripts, but it's not necessarily clear which one is the duplicate (this can depend on choices at that wiki). For these it's easy to add P460, but the next steps are more complicated. If a new property being used, this makes sorting them out much more complicated, unless it's symmetrical as well.
--- Jura 04:14, 31 March 2016 (UTC)- On nnwiki is it rather straightforward. The main script of the wiki is Nynorsk (nn). The duplicate articles are in Högnorsk, a third unofficial way to write Norwegian. If we ever will have a Högnorsk wiki, those articles will most likely be transferred to that wiki. In that way these articles are maybe not "permanent duplicates". But it m8 take decades before there is a Högnorsk wiki if ever. A little to big part of the Nynorsk articles are just copy-pasted Bokmål-articles (nb) that has not been translated yet. -- Innocent bystander (talk) 07:51, 31 March 2016 (UTC)
- I can't really say how it works at hywiki, just that they do have two and P460 links them. When I added them, I think quite a few items didn't have any statements at all. BTW, I think the templates to get additional interwikis work with P460. If we remove them, we would need to find another way.
--- Jura 06:15, 1 April 2016 (UTC)
- I can't really say how it works at hywiki, just that they do have two and P460 links them. When I added them, I think quite a few items didn't have any statements at all. BTW, I think the templates to get additional interwikis work with P460. If we remove them, we would need to find another way.
- On nnwiki is it rather straightforward. The main script of the wiki is Nynorsk (nn). The duplicate articles are in Högnorsk, a third unofficial way to write Norwegian. If we ever will have a Högnorsk wiki, those articles will most likely be transferred to that wiki. In that way these articles are maybe not "permanent duplicates". But it m8 take decades before there is a Högnorsk wiki if ever. A little to big part of the Nynorsk articles are just copy-pasted Bokmål-articles (nb) that has not been translated yet. -- Innocent bystander (talk) 07:51, 31 March 2016 (UTC)
- Comment I think this property label would be confusing and abused (as Jura seems concerned about) - what about something more specific: 'item referenced in alternate script' maybe? Is there any prospect of wikibase handling this sort of case better? ArthurPSmith (talk) 19:41, 1 April 2016 (UTC)
- The bulk on Wikidata:Database_reports/permanent_duplicates are articles in different languages (or dialects depending on the POV). The second item is really about the same as the first one. So "same as" isn't incorrect. It can be a different script, but it needn't. Maybe a symmetrically used "same as (WikiMedia permanent duplicates)" would do?
--- Jura 17:27, 3 April 2016 (UTC)- I updated it accordingly.
--- Jura 15:34, 6 April 2016 (UTC)- @Jura1: Hm, I'm not really in favor. I like making a firm distinction between the Wikidata page/item and the topic represented. Titling the property "same as", with extra disambiguating content, is misleading. There isn't topic X which is same as topic Y. The Wikidata item has, for technical purposes, an associated item here. The extra item has no topic, it's just a technical unit. --Yair rand (talk) 23:01, 11 April 2016 (UTC)
- Tricky. What solution do you suggest for the hywiki usecase presented above? Do you have some you came across and added statements? Maybe "items with sitelinks about the same".
--- Jura 09:51, 15 April 2016 (UTC)- For hywiki alternate script pages, only one of the two pages should be linked to the item that functions as the item for that topic. In my opinion, permanent duplicated items should have no statements beyond those identifying its status as a duplicated item of something else. "Same as" would be something we'd use for "real" items (ie, those with topics). --Yair rand (talk) 22:45, 1 May 2016 (UTC)
- The question is not what item should get what, but if items don't have any other statements, personally, I can only determine that they are the same, but not which one should get what. Thus the advantage of a symmetric property. BTW, I don't think they are alternate script pages.
--- Jura 07:27, 5 May 2016 (UTC)
- The question is not what item should get what, but if items don't have any other statements, personally, I can only determine that they are the same, but not which one should get what. Thus the advantage of a symmetric property. BTW, I don't think they are alternate script pages.
- For hywiki alternate script pages, only one of the two pages should be linked to the item that functions as the item for that topic. In my opinion, permanent duplicated items should have no statements beyond those identifying its status as a duplicated item of something else. "Same as" would be something we'd use for "real" items (ie, those with topics). --Yair rand (talk) 22:45, 1 May 2016 (UTC)
- Tricky. What solution do you suggest for the hywiki usecase presented above? Do you have some you came across and added statements? Maybe "items with sitelinks about the same".
- @Jura1: Hm, I'm not really in favor. I like making a firm distinction between the Wikidata page/item and the topic represented. Titling the property "same as", with extra disambiguating content, is misleading. There isn't topic X which is same as topic Y. The Wikidata item has, for technical purposes, an associated item here. The extra item has no topic, it's just a technical unit. --Yair rand (talk) 23:01, 11 April 2016 (UTC)
- I updated it accordingly.
- The bulk on Wikidata:Database_reports/permanent_duplicates are articles in different languages (or dialects depending on the POV). The second item is really about the same as the first one. So "same as" isn't incorrect. It can be a different script, but it needn't. Maybe a symmetrically used "same as (WikiMedia permanent duplicates)" would do?
- Support, although I would prefer a more specific name, possibly refering to Wikimedia projects. Of course, ideally I would prefer the ability to have multiple sitelinks to the same projects, as also required for Commons. --Srittau (talk) 20:05, 11 June 2016 (UTC)
Oppose I am not happy with the one-sitelink-per-project-and-item rule but adding this property would break a basic assumption of Wikidata data model. The problem could also be resolved in each single Wikipedia or file a bug report on Wikibase software. Could you give some more examples where it is actually needed? -- JakobVoss (talk) 12:27, 12 June 2016 (UTC)
- @JakobVoss: Which basic assumption is that (which would be broken)? Further examples: City of Brussels (Q239) => City of Brussels (Q13054517), Slavic studies (Q16634496) => Slavic studies (Q156864), Q17363906 => Estonia (Q191). These are currently using said to be the same as (P460), which has the unfortunate side effect of bringing the technical-necessity items into the scope of things-that-exist items, which has substantial potential for confusion and broken Wikidata queries. --Yair rand (talk) 22:53, 19 June 2016 (UTC)
- The assumption is to have one Wikidata item for each distinct conceptual entity. Otherwise there is no need to agree on statements and to work together. If tow items are same, they should be merged. If merging is not possible for technical reasons, like the examples above, file a bug report of feature request to Wikibase software. If merging is not possible for conceptual reasons, the items are not the same but only said-to-be the same (said to be the same as (P460)) -- JakobVoss (talk) 04:36, 20 June 2016 (UTC)
- I agree with you in principle. The current one-sitelink-per-item rule has not worked out and needs to be changed. This has been the source of all the woes with Commons integration as well. Unfortunately, the development team seems to have other priorities, even though I personally consider this a high-impact limitation as can be seen in this and other discussions. That said, I consider this proposal an appropriate stop-go measure. --Srittau (talk) 06:45, 20 June 2016 (UTC)
- For reference, I brought it up again here. --Srittau (talk) 06:58, 20 June 2016 (UTC)
- Whether we add this property or not, the duplicate items will continue to exist for the foreseeable future because it's not possible to merge them. If we later get support for multiple sitelinks on an item, we can use this property to find items which need merging and then delete the property. - Nikki (talk) 11:51, 20 June 2016 (UTC)
- The assumption is to have one Wikidata item for each distinct conceptual entity. Otherwise there is no need to agree on statements and to work together. If tow items are same, they should be merged. If merging is not possible for technical reasons, like the examples above, file a bug report of feature request to Wikibase software. If merging is not possible for conceptual reasons, the items are not the same but only said-to-be the same (said to be the same as (P460)) -- JakobVoss (talk) 04:36, 20 June 2016 (UTC)
- @JakobVoss: Which basic assumption is that (which would be broken)? Further examples: City of Brussels (Q239) => City of Brussels (Q13054517), Slavic studies (Q16634496) => Slavic studies (Q156864), Q17363906 => Estonia (Q191). These are currently using said to be the same as (P460), which has the unfortunate side effect of bringing the technical-necessity items into the scope of things-that-exist items, which has substantial potential for confusion and broken Wikidata queries. --Yair rand (talk) 22:53, 19 June 2016 (UTC)
- Support I think a dedicated property is a good idea. There's no obvious way to mark things as duplicates so people use various inconsistent methods. I think the label should start with "Wikidata" though so that it's more consistent with the other Wikimedia/Wikidata/etc properties/items. - Nikki (talk) 11:51, 20 June 2016 (UTC)
- Support I'm in favor, too. An advantage to having a property like this could also be that something might then be rigged up wherein the "permanent duplicate" is automatically fed the interwiki links of its twin all all languages not also having duplicates.
- For the record, on Ladino Wikipedia, the convention we are using is that the lad-latn version is always linked, while the lad-hebr version is by itself (or in some cases, especially template or project space pages, linked to other multiple version items). On the lad-hebr pages we use Module:Interwiki (Q20819069) to feed the interwikis from the lad-latn pages to the lad-hebr page. StevenJ81 (talk) 15:43, 20 June 2016 (UTC)
@Yair rand, Brya, Srittau, Jura1, Nikki, StevenJ81:
Done Now permanent duplicated item (P2959) -- Lymantria (talk) 05:36, 11 July 2016 (UTC)
- Thank you. @Innocent bystander: The module should now recognize this alongside (or even in preference to) said to be the same as (P460). StevenJ81 (talk) 12:54, 11 July 2016 (UTC)