Wikidata

Wikidata:Project chat/Archive/2017/12

< Wikidata:Project chat‎ | Archive‎ | 2017
This page is an archive. Please do not modify it. Use the current page, even to continue an old discussion.

sparql url

I am trying to open a sparql url like [1] but the last days is not opening. It is downloading a sparql url. Is only my problem? Did I make it happen? Xaris333 (talk) 20:34, 1 December 2017 (UTC)

Try this one. Please note https://query.wikidata.org/#. --Succu (talk) 20:41, 1 December 2017 (UTC)

Thanks! Xaris333 (talk) 10:36, 2 December 2017 (UTC)

This section was archived on a request by: Matěj Suchánek (talk) 19:44, 2 December 2017 (UTC)

Monolingual language code mn-Mong for traditional Mongolian entry in wikidata item.

About 1.5 years ago, some users have requested for the creation of monolingual language code mn-Mong on phabricator phab:T137810 so that Mongolian text in Mongolian script can be entered into wikidata. However, on phabricator, User:GerardM was saying that per policy of Language Committee, that code cannot be accepted because mn is a "macrolanguage" code. (macrolanguage (Q152559). As such, I would like to put the situation into discussion in Wikidata regarding what are we going to do with the code. My personal opinion is that the code mn-Mong would be the best fit for the values that are going to be added to wikidata despite what have been stated by Language committee, and as a result I believe the optimal solution is to establish a local consensus that would override the language committee policy.C933103 (talk) 19:42, 11 November 2017 (UTC)

@GerardM: How about mvf, given that this represents the variety spoken in Inner Mongolia and that variety is written in the Mongolian script? (If we wish to represent the variety spoken in Mongolia in the Mongolian script, we could use khk-Mong.) Mahir256 (talk) 22:32, 11 November 2017 (UTC)
The thing is, as far as I understand, if you write a text in the Mongolian script according to its orthography, it will be the same no matter if it is mvf or khk. In other word, if a Mong text is presented, you can't judge if it is mvf or khk. (That's what I was told by Mongolian speakers anyway)C933103 (talk) 22:35, 11 November 2017 (UTC)
@C933103: If that is true then why do you want mn-Mong instead of mvf? ChristianKl (✉) 22:43, 11 November 2017 (UTC)
They are not distinct when written in Mongolian script, but they are distinct vocally. The mvf only represent the vocal dialect. When a khk speaker write the language in Mong, they will not say they are writing it in mvf. C933103 (talk) 23:02, 11 November 2017 (UTC)
The core issue is that we seem to have a policy that sees ISO-639-3 as authority for what a language happens to be. mvf is a language according to that standard. Would it work for you as a compromize to create that language? ChristianKl (✉) 00:27, 12 November 2017 (UTC)
The way I see it is that, spoken language =/= written language, therefore while putting mvf, khk separately could arguably make sense in term of the spoken language, and also the classification would make sense when writing with latin or cyrillic alphabets that are more strictly phonetical representation of those variants, however it does not make sense when writing with Mongolian script.
Then about "compromise". I cannot speak for Mongolian community but my understanding is that it would defeat the purpose that Mongolian written in Mongolian text is supposed to be beyond the boundary of individual Mongolian subgroup. To khk speakers, the coding could also seems like giving away Mong to China, and alienating the community from the script.C933103 (talk) 03:14, 12 November 2017 (UTC)

No. There is no room for abusing standards. The language policies are fixed because the Wikipedia experience showed the fallacy of having "consensus" determined codes. We still suffer the consequences because there is no process to undo the damage. The process of adding codes to Wikidata must conform to the ISO-639 standard. The problem with the flood of codes for Wikidata is that there is too much of them and no thought on what to use it for. At this moment it is used for text only fields. So when there it is about languages that is only spoken, by definition the use of such codes are problematic. I have asked at previous request for an argument for the codes, I did not get it and as far as I am concerned there are too many requests and the case for their creation is not made. Thanks, GerardM (talk) 07:28, 12 November 2017 (UTC)

That is immaterial. There is only one language policy and it is valid for any and all Wikimedia projects. Thanks, GerardM (talk) 08:52, 12 November 2017 (UTC)

Monolingual language code creation process

  • Wikidata seems to use language codes not only to talk about languages but also to talk about dialects. GerardM seems to have decided in https://phabricator.wikimedia.org/T151186#2916575 against having en-US as dialect within Wikidata. The next time the request for an opinion on https://phabricator.wikimedia.org/T154589 was ignored. At https://phabricator.wikimedia.org/T151186 there's another request. I can understand that the jurisdiction for deciding which new Wiki's are supposed to be oppened is at the language committee but I don't really see why the jurisdiction for Wikidata languages/dialects shouldn't made inside our community but outsourced to the committee. ChristianKl (✉) 18:23, 12 November 2017 (UTC)
    • It could be good to have a group of users interested in these questions reviewing them, but they'd need to collaborate with Wikidata. Currently I don't really see this happening. At some point, we got the explanation that the committee lost interest in reviewing these requests.
      • No explication is given why phab:T151186 took 4 months for an opinion and no explicit approval was given. The result was that this simple request took almost a year.
      • At phab:T165648, the request seems to be taken hostage by the language committee because there is some disagreement about a related language code currently used by a project grouping several languages. The language codes requests are considered correct by everyone involved and no other code is used at WMF for the same. There is no technical problem to use the code at Wikidata for monolingual strings.
    • Given that the Wikidata approach works fairly well for creating databases and Wikidata administrators are fairly proficient in building things that way, maybe Wikidata administrators would be a better group to review these requests. We do need a group that is interested in reviewing these requests and collaborates with Wikidata.
      --- Jura 08:35, 13 November 2017 (UTC)
  • Processwise we could do it similar to property creation. We can have one WikiProject that gets pinged, so people who are interested in preventing abuse of codes can voice their opinion. If the process would be successful, instead of creating a new property the admin/property creator could write a phabricator ticket that links to the language tag creation proposal.
@LydiaPintscher, GerardM: Would any of you have a problem with switching our process in this way? ChristianKl (✉) 12:47, 13 November 2017 (UTC)
From my side what I need is some confirmation that the language code we are adding isn't bogus/wrong/... because I don't have the necessary knowledge and time to judge this. So if this process leads to that I am fine with it. --LydiaPintscher (talk) 18:21, 13 November 2017 (UTC)
The process for properties is broken; it happens all too regular that properties are deleted. At a time adding languages was a community process and that was a failure. It is still not possible to undo the damage. We are not bringing a community process back for Wikidata. When monolingual codes were introduced, I was in favour of this. As it was discussed at the time. When the content for a code is wrong all its content is to be deleted. The idea is that this is to help not hinder.
The problem with many of the later Wikidata codes is that no reason is given for their creation. The fact that these codes are introduced regardless sets them up for failure and deletion. Thanks, GerardM (talk) 06:04, 14 November 2017 (UTC)
  • @GerardM: Given that you were till now either unwilling or unable to provide any examples of harm, I'm not sure why I should put much stock into your argument. It also makes me question to what extend the language committee is well equipped for their task when they don't maintain easily sharable lists of problems that they see. ChristianKl (✉) 13:10, 14 November 2017 (UTC)
  • I can't judge if language codes are valid or not. I'm only involved because I know how to add them to the code. If languages are accepted it's easy for me to make a patch. Mbch331 (talk) 16:59, 14 November 2017 (UTC)

Maybe the best way would be to still consult the LangCom but only for advices, for having a recommendation point of view and to consider that silence implies consent if their is no answer after some time (eg. 1 month). Cdlt, VIGNERON (talk) 11:24, 21 November 2017 (UTC)

There can be no doubt that there is no silence. There can be no doubt that I have said enough times that a macro language does not identify a language and consequently cannot be used. The harm involved is that something is implied that is not correct, something that does not conform with the standard used of the what the standard stands for. Just ignoring this does not change one whit why it is wrong on first principles. Thanks, GerardM (talk) 12:22, 21 November 2017 (UTC)
GerardM You are talking about one specific case, but we are talking about the general process (as indicated on Help:Monolingual text languages). And I'm not saying that you are silent (the langcom was indeed silent in many case, or late to answer, but in the other hand was not silent in many other cases, see phabricator:T144272 for a recap), I'm saying that in case of silence we should pursue without waiting for the langcom. Cdlt, VIGNERON (talk) 14:45, 21 November 2017 (UTC)
There is just no use to wait for people who seem to have lost interest. We still have no plausible explanation for the delays in "fr-ca", "nrf-je". Obviously, anybody can give valuable input.
--- Jura 19:09, 21 November 2017 (UTC)

mn-Mong (cont)

  • There is no inherent problem with en-US, the same is true for Australian English. There is one big problem. It is that liberties are taken with the meaning of the codes. A great example is when macro codes are used in combination with other codes like with Mongolian. What I find is that many codes are asked to be recognised without any argument why they are requested. There is no point in asking for a code for a language that is not written for instance. The point of the language policy is that the codes are used in the way they are meant to be used. We are still suffering the consequences from the abuse we have seen in the past. Thanks, GerardM (talk) 12:17, 13 November 2017 (UTC)
  • I think it would great if there's documentation of "suffering the consequences from the abuse we have seen in the past". Documenting the problems might help with increasing their visiblity, which inturn helps with solving them.
When it comes to this Mongolian example it seems to me that if we would use ISO-639 codes, then we have to give two language with distinct verbal expression but a shared written language different codes. Given that our project only uses the written language I don't see why this would be a problem. ChristianKl (✉) 12:47, 13 November 2017 (UTC)
It is a problem that "we" give a code that is external to us a meaning it does not have. Your notion of language is at odds with the standards. Thanks, GerardM (talk) 14:55, 13 November 2017 (UTC)
What do you mean by giving it the meaning that it does not have? The combination have already been registered on IANA Language subtag registry, and French Wikipedia have already been using Module:mn-Mong for a number of things already. C933103 (talk) 19:45, 13 November 2017 (UTC)
Why does it matter what the French Wikipedia does. It is wrong and not acceptable. Thanks, GerardM (talk) 06:05, 14 November 2017 (UTC)
More relevant is that Amir Aharoni wants all the language codes to be merged. This makes these often arbitrary request even more dangerous. There are plenty of codes that are fine but codes like mn-Mong are not acceptable not only because of the inclusion of an ambiguous code but also because it is given a meaning it does not have. Thanks, GerardM (talk) 07:21, 14 November 2017 (UTC)
Where have you seen that "mn-Mong" is wrong ? It does not indicate the country (irrelevant here) but the traditional Mongolian script (traditionally written vertically) used for the Mongolian language (modern Mongolian generally uses Cyrillic in Mongolia, but in China it uses now sinograms; the traditional script was used in what is now in Mongolia, part of Russia, and part of China: "mn-Mong" is valid in the three countries and refers to this orthograghy. "mn-MN" would likely be written now only in Cyrillic but would be only the more standard modern use in Mongolia, where "mn-Cyrl" may be used also in Russia and China for the same Cyrillic version. "mn-CN" would likely be written now only in simplified sinograms but would be only the more standard modern use in China, where "mn-Hans" would be used in the 3 countries without ambiguity.
In general we should avoid language codes using ISO3166-1 country/territory extension: languages are better encoded using ISO 15924 script codes, or variant codes for regional dialects.
So "mn-Mong" is far better than "mn-MN" (not needed: use "mn" instead and preferably use the cyrillic script only), "mn-CN" (use "mn-Hans" instead), "mn-RU" (use "mn" instead and only the cyrillic script). "mn-Cyrl" is not needed for Mongalian names in Russia or Mongolia (use "mn" directly) but "mn-Cyrl" could be useful in for names used in China. "mn-Latn" would also be useful to show romanizations (preferably from the Mongolian script if available, otherwise from the cyrillic version, otherwise from the sinographic version). There may be also useful cases for "mn-Hant" (traditional sinograms, used for historical name in the Inner Mongolian region of today's China before its invasion by Han imperial armies, but also previously when Mongols where invading today's northeastern China). But there are certainly regional variants of Mongolian which may also be written in several scripts.
Note: all these variants are not needed for the creation of a new wiki. This is purely for translations of Wikidata labels which should adequately identify languages or dialectal and orthographic differences rather than regions. Verdy p (talk) 17:30, 14 November 2017 (UTC)
Another note: being a "macrolanguage" does not mean we cannot encode data. Chinese (zh) is ALSO a macrolanguage, in which standard Mandarin is only one member. But because this Mandarin language is highly predominent, it was decided to alias "zh" and "cmn" together (also because the simplified script is now predominant, "zh-Hans" is aliased in IANA registry for BCP47 by making "Hans" the default script. There's evidence that Mongolian also has a predominant language variant, the one used today and standardized in Mongolia (with the Cyrillic script).
Macrolanguages are not blockers at all: a macrolanguage is distinct from a language family in that it has a large mutual understanding with minor differences that are easy to understand
(at least orally: the script however is hard to decipher for most native speakers even if this is exactly the same oral language within the macrolanguage group).
This is not the same about language families (notably "qu" for Quechua for which a more precise ISO 639-3 individual language code should be used as there's no evident mutual understanding and differences are no longer minor as they were in their old origin where they may have first emerged as dialects then as a macrolanguage, before being split like it occured between "ht" and "fr" when it emerged as a vivid creole, or between "fr" and "gcf"; as well the modern Filipino is first started as a dialect, then as a creole with good understanding i.e. a macrolanguage, then was separated definitely from the traditional Filipino dialect and today's Filipino is a strong creole separated from its 3 roots: old Filipino, Spanish and English...). Verdy p (talk) 17:48, 14 November 2017 (UTC)
@GerardM: So, what do you mean by "giving it the meaning that it does not have"? C933103 (talk) 21:39, 15 November 2017 (UTC)
What is written above is about interpreting, giving it a meaning and it is not how we should deal with it. Thanks, GerardM (talk) 05:44, 16 November 2017 (UTC)
The IANA language tag registry does define criteria for what they allow as languages and follow the IETF standards. I see no reason why we need to follow the ISO standards here when the IETF seem more useful. ChristianKl (✉) 14:00, 14 November 2017 (UTC)

Why not learning how Azerbaijani (Q9292) works?

The Azerbaijani is also a macrolanguage and have both North Azerbaijani (Q3515311) and South Azerbaijani (Q3449805), so we can assign both as "az" and "azb", therefore I suggest continue using "mn" for things in the Mongolia country, and "mvf" for Inner Mongolia, why not? --Liuxinyu970226 (talk) 13:24, 15 November 2017 (UTC)

The distinction between Mongolia country/Inner Mongolia seems to be about regions and not about scripts. People in Mongolia country can still use the traditional Mongolian script even when Cyrilic script is the default for Mongolia country. ChristianKl (✉) 13:38, 15 November 2017 (UTC)
The language policy is applicable for new projects. The code for az predates it. We would not have an az.wikipedia.org if the decision was to be made today.
Your understanding of language codes proves to be problematic. The code mn is for everything Mongolian. Any other interpretation is wrong. Thanks, GerardM (talk) 15:30, 15 November 2017 (UTC)

"Any language allowed in Wikidata"

According to this WMF blogpost, seems like langcom have reached some consensusn back in 2013. It mentioned that "Languages used with multiple scripts need to be configured in this way.". What does it mean? Also, the post mentioned requirement for ISO 639-3 code but I don't see any point saying ISO 639-3 macrolanguage codes being excluded.C933103 (talk) 05:43, 17 November 2017 (UTC)

When a language is written in multiple scripts, it is a combination of the language code followed by the script code.. eg sr-Cyrl or sr-Latn. For whatever reason we do not include the capital on the script code. Thanks, GerardM (talk) 18:26, 21 November 2017 (UTC)
So, why are macrolanguages being excluded? It's said on phab that that mean multiple languages, but they are grouped together as macrolanguage for reasons. C933103 (talk) 18:35, 27 November 2017 (UTC)
Because a string of tekst is in one language not multiple languages (Obvious really). Thanks, GerardM (talk) 17:19, 28 November 2017 (UTC)
If you look at http://www-01.sil.org/iso639-3/scope.asp#M then you can see a use case for those macrolanguage code to be established is that, they have a "common written form used for multiple closely-related languages". Therefore, for the purpose of writing an online encyclopedia or writing entry in a database, which are not spoken languages, I don't see why there would be a need to break down the code into specific individual languages. @GerardM: C933103 (talk) 10:11, 1 December 2017 (UTC)

Wikidata weekly summary #288

 
Here's your quick overview of what has been happening around Wikidata over the last week.
  • Development
    • Result views in the Wikidata Query Service UI can now be configured with options (phabricator:T155973, Special:Diff/580233265/596922582)
    • Worked more on introducing the concept of sub.entities in the APIs in order to make edits to Forms persistent
    • Worked on edit summaries for edits to Forms

You can see all open tickets related to Wikidata here.



Something that looks sad for me: We have a consensus to deprecate original language of film or TV show (P364) but nothing is proceed after that: Wikidata:Properties_for_deletion#P407 and P364. This clear told me that the consensus can also be failed. --Liuxinyu970226 (talk) 11:22, 1 December 2017 (UTC)

Find Items with number of statements!

At WIKIDATACON, it was mentioned that eventually items could be found, based on quality! But, until then;

Is there a way to find/list/query items with the number of statements it contains?

and, items with no labels in a particular language? or items with only one label, with no additional languages?

and, items that have no "instance of and subclass of" statements? (without using the game!)

and, items with no external identifiers?

If it has already been answered!? I apologise for asking again! – The preceding unsigned comment was added by Wallacegromit1 (talk • contribs) at 7. 12. 2017, 04:29‎ (UTC).

  Comment moved to WD:RAQ
This section was archived on a request by: Matěj Suchánek (talk) 17:20, 7 December 2017 (UTC)

166 duplicates - municipality of Venezuela

At

the 166 items from Q31961198 to Q32491773 probably all duplicates from cebwiki. The Swedish botmaster seems to not have linked the cebitems in WD 92.226.219.66 16:49, 28 November 2017 (UTC)

User:GerardM - can you help with merging? Spanish name is "Municipio X" , English name is "X Municipality". The correct state must be selected, example for ambiguous name : en:Arismendi Municipality. 77.180.4.196 09:53, 29 November 2017 (UTC)

I have another project tat is a time sinc. thanks, GerardM (talk) 12:05, 29 November 2017 (UTC)

What's up with the duplicates? 92.227.88.124 12:05, 2 December 2017 (UTC)

@Mr. Ibrahem: Your bot created ca. 166 duplicates, can you merge the items? 77.180.168.221 12:10, 2 December 2017 (UTC)

Merging male (Q6581097) with male organism (Q44148)

See the discussion site of male (Q6581097). ChristianKl (✉) 03:22, 2 December 2017 (UTC)

Send a 'thanks' notification to one contributor for a log entry: it is almost possible...

Hello,
Have you ever wanted to send a 'thanks' notification to one contributor for a log entry? It can become possible... if you add your support vote to this proposal in the 2017 Community Wishlist Survey !
Regards --NicoScribe (talk) 09:27, 2 December 2017 (UTC)

Pasleim - PLbot nonsense edits - municipality of Venezuela

Once more Mr. Pasleim ... https://www.wikidata.org/w/index.php?title=Q6063601&diff=210151732&oldid=210151723  – The preceding unsigned comment was added by 92.226.219.66 (talk • contribs) at 28. 11. 2017, 17:43‎ (UTC).

I don't think the project chat is the right place for such exposures. In case you want to have me blocked, start a discussion at WD:AN, in case you want to have me globally banned (like Tamawashi) follow the steps at m:Global bans. --Pasleim (talk) 17:54, 28 November 2017 (UTC)
User:Pasleim - "I don't think the project chat is the right place for such exposures" - you are a mass (mis) contributor and maybe items from other areas of interest have become victim of your mis-edits. 78.55.243.85 18:41, 28 November 2017 (UTC)
It isn't obvious to me why that is nonsense. Please change your tone otherwise it will be you who gets blocked. You might feel being important but you're just annoying. Matěj Suchánek (talk) 19:10, 28 November 2017 (UTC)
It's User:Tamawashi... --Succu (talk) 19:42, 28 November 2017 (UTC)
It is Tobias Conradi.--Ymblanter (talk) 21:22, 29 November 2017 (UTC)
Who else? Matěj Suchánek (talk) 19:40, 2 December 2017 (UTC)

User:Matěj Suchánek "It isn't obvious to me why that is nonsense" - what don't you understand? 92.227.88.124 12:07, 2 December 2017 (UTC)

I don't understand it as well. What is the problem? --MB-one (talk) 18:08, 2 December 2017 (UTC)
You have shown a link to a diff where a bot added a valid statement... I really wonder what is wrong. You just want to be visible, nothing more. Matěj Suchánek (talk) 19:40, 2 December 2017 (UTC)

von Surname and van Surname

When we use surname categories is it "van Gogh" or "Gogh"? I tried a few vans and vons and surnames had one of each type, so what should we harmonize on? In early times everyone was from somewhere before there were surnames. Later they become defacto surnames.--RAN (talk) 02:12, 1 December 2017 (UTC)

Well, it is the Van Gogh Museum, not the Gogh Museum. - Brya (talk) 04:11, 1 December 2017 (UTC)
It clearly is "Van Gogh" or even "Vincent van Gogh". Other than you suggest such names did not become de facto surnames, in the Netherlands people had to choose an official surname in the Napoleontic era. Lymantria (talk) 06:26, 1 December 2017 (UTC)
This is a complicated issue - there are rules but there are no universal rules, and it varies between countries. We shouldn't try and force a harmonisation. Andrew Gray (talk) 12:31, 1 December 2017 (UTC)
Dutch family names contain tussenvoegsels. I think we should have a seperate property for that, because the family name is "Gogh" (also sorted on that). Sjoerd de Bruin (talk) 14:25, 1 December 2017 (UTC)
Mostly, Dutch names are sorted on "Gogh", but not always. On the other hand Belgian names are sorted on "Van" (capitalized by the Belgians). - Brya (talk) 04:43, 3 December 2017 (UTC)

Wikidata list

Anyone knows why is not working? User:Xaris333/Test. The query is correct [2]. Xaris333 (talk) 20:06, 2 December 2017 (UTC)

I have the same question for https://www.wikidata.org/wiki/Wikidata:WikiProject_Properties/Reports/Missing_Examples ChristianKl (✉) 20:17, 2 December 2017 (UTC)
(edit conflict) The query shouldn't contain "|", you need to rewrite it. Matěj Suchánek (talk) 20:22, 2 December 2017 (UTC)
I think that Listeria work only with distinct item and query don't return item --ValterVB (talk) 20:20, 2 December 2017 (UTC)
I tried several things among this one, but it did not work. If one outputs wikibase:Statements in ?item (which shall be listed here) [3], Listeria apparently cannot handle it. Pretty sure now that this it does not work at all with Listeria. —MisterSynergy (talk) 20:28, 2 December 2017 (UTC)
It doesn't work here too Wikidata:WikiProject Properties/Translation, probably because it uses wikibase:Statements. John Samuel 20:32, 2 December 2017 (UTC)
Listeria need that the query return item (Qxxxx), all the other fields are: or SPARQL variable "?zzzzz" or direct claim of the item. No chance to have something of different --ValterVB (talk) 20:38, 2 December 2017 (UTC)

Oxford Biography Index Number (P1415)

Property:P1415 the scheme has changed with no redirect from the old scheme to the new scheme. Click on the example in the property entry. --RAN (talk) 16:44, 30 November 2017 (UTC)

Going to be a bit of an issue as there is not a one for one replacement, unless they have allowed some magic

Bathurst, Henry (1781–1844)

OLD

NEW

So we can swap in the url fragment http://www.oxforddnb.com/view/10.1093/ref:odnb/9780198614128.001.0001/odnb-9780198614128-e- however the 101001697 differs from 1697

Someone is either going to do some testing, or contact them.

As they have DOIs available, I would suggest that we would be better to convert to those, and fix the items at the same time. That should have a longer shelf-life.

https://doi.org/10.1093/ref:odnb/1697

 — billinghurst sDrewth 17:41, 30 November 2017 (UTC)

What a pain. Why do sites do this, without providing a migration route?
Do we know what yet they've done yet with articles covering multiple people? Previously, it could have several identifiers, one per person. If they've moved over to DOIs, does that mean only one DOI per article.
Also, previously the Oxford identifiers were intended to cover not just the ODNB, but in some cases also other Oxford resources for the same individual, linking them together. Does this mean they've abandoned that baby-steps attempt towards authority control across their different resources? Jheald (talk) 09:31, 1 December 2017 (UTC)

@Charles Matthews, Magnus Manske, Richard Arthur Norton (1958- ), Billinghurst, Jheald:. I did a bit of work on this last night. In short:

  • The old ID forms we used were "Oxford Biography ID Numbers", which were a padded and extended form of the shorter ID used internally by the ODNB. 16977 became 101016977. This went to a landing page rather than the article directly, which included the ODNB plus any entries elsewhere in Oxford resources (Who's Who, Grove, etc)
  • The new IDs are based around the old shorter form of the ID - back to 16977 - and go straight to the target article.
  • "Child" identifiers for a second entry in the same article - 16978 - are still valid, but redirect to the "parent" URL with a # anchor to the relevant section. The DOI for the second entry isn't displayed on the page anywhere.
  • The old "Oxford Biography ID Numbers" seem to have vanished entirely, unfortunately (which is perhaps not surprising - they were never really used for anyone not in the ODNB). In practice, we always used P1415 as primarily for the ODNB not the other resources.
  • It is fairly trivial to generate long forms and short forms of the ID from each other

My plan is to:

  1. pull out a list of all currently live P1415s
  2. delete the existing P1415s
  3. change the formatter URL to use the DOI form (10.1093/ref:odnb/$1)
  4. reimport P1415 with the "short" forms
  5. tidy up P1415 documentation to focus it on the ODNB rather than the OBIN
  6. poke around to fix any wikis which are auto-importing this

This will take a couple of days and during that time perhaps it would be best to disable the ODNB formatter URL to avoid confusion.

Thoughts? Andrew Gray (talk) 12:15, 1 December 2017 (UTC)

I agree with switching to the DOI form: this was done a while back on enWP, which therefore is faring better. Thanks for undertaking the transition. Charles Matthews (talk) 16:32, 1 December 2017 (UTC)
  • Excellent! While we are on the subject, does anyone have access to a paid account there, so I can get a copy of an entry? --RAN (talk) 00:26, 2 December 2017 (UTC)
    • Most of us in the UK have online access through our local public library cards -- the local libraries mostly have full subscriptions, that allow offsite online use. Jheald (talk) 20:43, 2 December 2017 (UTC)
  • @Andrew Gray: Thanks for taking this on. What you suggest sounds a very good plan -- let's hope most or even all of the short forms are still working. Do we have any idea whether new short forms will be minted for the new entries on 1 January 2018? Or will the latest cohort be the last? Jheald (talk) 20:43, 2 December 2017 (UTC)
  • The "short forms" are the numbers they've been using in their DOIs and most URLs all along - the OBIN was a bit of a weird roundabout approach on our side to get the landing page and any other items linked from it. I don't anticipate any changes in this approach so we should be good to standardise on the short form.
On that note, I have it more or less ready to go in an enormous quickstatements batch - if there are no objections I'll start running the changeover tonight. Andrew Gray (talk) 20:51, 2 December 2017 (UTC)
Okay, going! Will leave it running overnight. Andrew Gray (talk) 22:40, 2 December 2017 (UTC)
  •   Comment Maybe it should be said that in general one should avoid changing property values. The approach would usually be to create a new property and add the new values there.
    As P1415 is probably only marginally useful, I suppose not much harm is done in this specific case.
    --- Jura 08:31, 3 December 2017 (UTC)

Property for a period

What property can I use to describe the historical or archeological period or era an item belongs to? Breg Pmt (talk) 11:46, 2 December 2017 (UTC)

The property that's named time period (P2348). It's surprising to me that you didn't find it. Is there some user interface issue that made it hard for you to find it? ChristianKl (✉) 12:43, 2 December 2017 (UTC)

@ChristianKl: Thank you. It was hidden under an unusual Norwegian item name and no description. (and to me also personally the English was rather unexspected.) Breg Pmt (talk) 12:48, 3 December 2017 (UTC)

@Pmt: Then I recommend to give it a better Norwegian name or add aliases so that the next Norwegian is going to find it. ChristianKl (✉) 18:28, 3 December 2017 (UTC)

Wikidata:Open data publishing

Hi all

I've produced a page aimed at providing data producers and publishers with basic information about open data, database rights and publishing best practices. My aim with this resource is to improve data producers understanding to where they can make informed decisions about the theory and practicalities of open data publishing.

There are many other organisations e.g ODI and OKI who are already producing excellent content so I kept the page as a brief overview and link out to more in depth explanations on each subject.

I think there is some additional work needed on the page, especially around:

  • Additional guidance on choosing data licenses.
  • More options and guidance around best options for data publishing platforms.
  • The data formats in the table needs some work, I think you can have 5 star data simply by linking in a CSV file to Wikidata items rather than using special formats? This seems very important for smaller organisations with less resources.
  • I'm not sure the page name is correct, any ideas?

Additionally I think it would be helpful to make the information on database rights available on other pages on Wikidata as well. There seems to a lower level of understanding of database rights compared to copyright in Wikimedia in general. Can anyone suggest places it could be added?

Thanks

John Cummings (talk) 13:51, 3 December 2017 (UTC)

@John Cummings: The website looks nice. One issue that it doesn't talk about is the difference between licenses. It would make sense to add a section about publishing data as CC0 being the best practice. ChristianKl (✉) 15:10, 3 December 2017 (UTC)
Hi ChristianKl, thanks, do you know of any external sources that recommend CC0 for data as best practice for data publishing? --John Cummings (talk) 16:22, 3 December 2017 (UTC)
Unfortunately, I don't have good links for you. Basically the US government created data is public domain by default. The fact that the German National Library uses CC0 for the Integrated Authority File (GND) might be worth expressing. ChristianKl (✉) 16:26, 3 December 2017 (UTC)

problem with Special:Watchlist & Wikidata is slow

Tracked in Phabricator
Task T182318
Resolved
Tracked in Phabricator
Task T182322
Resolved

This morning, Special:Watchlist just fell back to old interface... see screen capture here.

The improved display, like on Recent changes, ceased to work... as the grouping of diffs by page... what happened ?

Also, only the last edit to a page is displayed, which means that if the last edit is mine, there is no display of the page at all, like for the "Project chat" that is not displayed on the capture... --Hsarrazin (talk) 10:35, 7 December 2017 (UTC)

Also, the web interface is unbearably slow today. It's practically unusable. --Anvilaquarius (talk) 12:07, 7 December 2017 (UTC)

Hello,
Thanks for noticing. We're looking at these two things as soon as possible. Lea Lacroix (WMDE) (talk) 12:09, 7 December 2017 (UTC)
It's still unworkable slow. Apparently after 6 december 23:00. See also grafana. Is something changed since then? Rudolphous (talk) 10:44, 8 December 2017 (UTC)
We're still investigating but it seems other sites are affected as well. Wikidata is just hit especially hard because a lot of our user interface relies on the API. Updates will go into the ticket as we make progress. --Lydia Pintscher (WMDE) (talk) 11:25, 8 December 2017 (UTC)
Thanks for the rollback (Q395307). All is fine now again. --Succu (talk) 23:13, 8 December 2017 (UTC)
Thanks to all for resolving the error. It seemed to be quite tricky to find out what the problem was, judging from the Phabricator discussion. --Anvilaquarius (talk) 11:19, 9 December 2017 (UTC)
This section was archived on a request by: Matěj Suchánek (talk) 11:24, 9 December 2017 (UTC)
works indeed like normal again. Also thanks from me! Rudolphous (talk) 19:16, 9 December 2017 (UTC)

2017-11-20 vandalism by IP - no one cared until I came

https://www.wikidata.org/wiki/Special:Contributions/189.168.18.255 92.226.213.195 14:16, 7 December 2017 (UTC)

Thank you, that user is no longer active. Bigbossfarin (talk) 15:50, 7 December 2017 (UTC)
This section was archived on a request by: Matěj Suchánek (talk) 12:23, 9 December 2017 (UTC)

Person vs. occupation

Hi!

I am a beginner on Wikidata, and I would like to understand it better. I just checked the Q2526255 (film director), and I saw, that the English and German description shows, this is a person, while the Hungarian shows, this is an occupation. The first statement is: „instance of” „filmmaking occupation”. Its subclass is director, which shows this is an occupation, statement shows this is a profession and a position. As I saw, this items are used in the personal pages (items with persons) in the occupation statement.

  1. May I change the descriptions of the film director item so, that this is an occupation (and not a person)?
  2. How can I decide if I should use the "occupation" or the "profession" or the "position" values in the instance of statements? Is the best way to use all of these?
  3. Is there a list about the statements and their values, or is the best tool the "What links here" (for example)?
  4. Is there a documentation where it is described, for which cases and situations which statements should we use, as OSM does with the tags (example)?

Thank you for your answers. Samat (talk) 09:52, 30 November 2017 (UTC)

Short answer: an occupation is not a person, so no. Longer answer: it occurs quite often that articles are added to the wrong items in a mistaken attempt to preserve interwikilinks. This appears to be the case with the Hungarian articles linked to film director (Q2526255) and filmmaking (Q932586). I would always assume the big 5 (en/de/fr/ru/ja) to be more correct than little languages (like hu). Jane023 (talk) 10:18, 30 November 2017 (UTC)
@Jane023: thank you for your answer, but I think, I was not clear enough. We are using everywhere professions, like film director (Q2526255) or director (Q3455803) as occupation in the items of persons (see examples in Tom Hanks (Q2263) or Stephen King (Q39829)), and the items of these professions also claims, that they are an occupation (first statement is "instance of" "filmmaking occupation" or "profession" or "position"). I think, the small language (hu) is the correct one in this case. Samat (talk) 10:32, 30 November 2017 (UTC)
I was referring to the label/descriptions. It is easier to adjust the sitelinks of certain languages than to reverse usage of an item in many other items. Please take the path of least resistance. Jane023 (talk) 11:15, 30 November 2017 (UTC)
To the sitelinks: I didn't check them earlier, because I wanted to clarify the Wikidata ontology and practice first. Now I checked them, and as Fractaler mentioned below, there is complete chaos here: most languages I can understand present the person and the occupation in the same article. In spite of the article starts with a statement, that this is a person, it describes the occupations, and categorize them as an occupation. Same as here, in Wikidata. To the path of least resistance: I understand your argument, but I don't agree with it, because this is an optimum only for short-term. If there is a mistake in 9 from 10 items, I would rather correct the nine article today, so we can reach the ordered state. Last year Wikidata has around half of the items it has today, and probably next year it will have double the number of items it has today. If the new items will be inserted into a clear and correct system, the work we have to do is much smaller on long term. For example if in a few years we have 10x larger database, than we have 100 correct items and for that we had to correct only 9 at the beginning. The other option is GBs of useless mess. Samat (talk) 19:01, 30 November 2017 (UTC)
This is not a Wikidata problem, and it is not your problem. Most of these sitelinks came in with the original 2013 "great interwikilink migration". You can pick up a mess and work away at it if you want to, but please be aware you may find certain Wikipedias objecting. It is up to those communities to link their articles to the correct items and not the other way around. Jane023 (talk) 17:31, 1 December 2017 (UTC)
The way it is used strongly suggests that the item is about the concept of an occupation (or position). Presumably the descriptions need to be adjusted. - Brya (talk) 11:41, 30 November 2017 (UTC)
Items, which have "person + occupation" (director (Q3455803), Q41515083, etc.) are homonym (Q160843) (=chaos (Q1787424) in Wikidata as knowledge base (Q593744)). Fractaler (talk) 11:12, 30 November 2017 (UTC)
I see :) Isn't there a conception or idea to separate the person and the occupation in Wikidata? One for instance of human and an other for instance of occupation? Samat (talk) 19:01, 30 November 2017 (UTC)
The thing is that Wikidata editors (Q28859214) (as superorganism (Q916139)) at the moment has some member of a group (Q36809769) that do not have Q4329817 and this is not eliminated. Therefore, it is very difficult to put things in order through direct editing (for example, by entering more precise items instead of homonyms, separate the person and the occupation in Wikidata: director -> director (profession), director (position), director (human)). There remains the only way - enlightenment (ie, wait until the person himself realizes that he was wrong). --Fractaler (talk) 07:39, 1 December 2017 (UTC)

Occupation vs. profession

Sorry to copy the questions from above, but I have the experience, that every section should contain only one question, others will be neglected. As it happened now. And I try with 3 questions again :)

  1. How can I decide if I should use the "occupation" or the "profession" or the "position" values in the instance of statements? Is the best way to use all of these?
  2. Is there a list about the statements and their values, or is the best tool the "What links here" (for example)?
  3. Is there a documentation where it is described, for which cases and situations which statements should we use, as OSM does with the tags (example)?

Thank you in advance. Samat (talk) 19:01, 30 November 2017 (UTC)

We should have some best practices at Wikidata:WikiProject Occupations and professions, but it's very confusing still. I believe the distinction is "profession" for things for which a person is formally educated (doctor, lawyer, computer scientist). "Occupation" is for what they actually *do* (hospital administrator, politician) and includes hobbies and non-"professional" jobs (singer, knitter, activist). I think there is still a lot of confusion around positions that require a formal apprenticeship or other expertise. "Position" is a role held by the person - so a US senator might have profession=lawyer, occupation=politician, position=senator. I'd really like to see a discussion of how to use these properties so we can be more consistent. I think some of the confusion is that the nuances of these terms are different in different languages (and even among English speakers, where the older definition of "the professions" is really not well understood anymore). - PKM (talk) 19:27, 30 November 2017 (UTC)
Dear PKM, thank you for your detailed explanation. This sounds clear and make sense for me. But in practice it is not so easy many times. :) For example somebody can be a singer with or without a formal education, and we cannot decide on the singer page which is the case. (And the singer (Q177220) item uses professional instead of occupation ;)). Samat (talk) 20:11, 30 November 2017 (UTC)
Yes, singer (Q177220) is also homonym (Q160843) (earlier street singers. for example, had no education). --Fractaler (talk) 07:39, 1 December 2017 (UTC)
So, 1 item (1 homonym) "US senator" -> 4 items (4 non-homonyms): 1) US senator (profession), 2) US senator (occupation) US senator (position), US senator (human). --Fractaler (talk) 07:44, 1 December 2017 (UTC)

To my next to last question: I found the list of properties, and the values of a statement are not restricted, so cannot be listed, if I am right. Is there any good answer to my last question? Samat (talk) 20:11, 30 November 2017 (UTC)

No, there are no consensus-based guidelines on when to use each item. In fact profession (Q28640) is currently defined as both <subclass of> occupation (Q12737077) and <different from> occupation (Q12737077), which makes no sense to me, although I understand trying to make a distinction. It doesn't help editors that "profession" has the alias "occupation" in English. At some point, we should start a project to create guidelines and best practices for these terms, but it's currently very messy and I am not sure we could reach consensus. - PKM (talk) 21:06, 3 December 2017 (UTC)

Mix n' Match matching tips

Hi all

Since a lot of us use Mix n' Match to import data and want the matching to be as good quality as possible I've started a 'matching tips' section on the manual to help matchers make good decisions when matching, please take a look and add additional ones. Thanks to everyone who gave their two cents in the earlier discussion about this.

https://meta.wikimedia.org/wiki/Mix%27n%27match/Manual#Matching_tips

Thanks

--John Cummings (talk) 19:37, 3 December 2017 (UTC)

Sources oppose to qualifier

How to express that? For example, all agree that state of Barcelona (Q1492) was Spain till 2017. But Catalonia states that this is valid till 2017 and Spain states this is valid without end date. How to express this better? To have 2 values "Spain"? Similar question with Iturup (Q215727) and other Kuril Islands (Q80389) where the date of Japan's ownership is under discussion. --Infovarius (talk) 19:53, 3 December 2017 (UTC)

And how to add a source to specific qualifier (e.g. one source for beginning date and the other for ending date)? --Infovarius (talk) 19:55, 3 December 2017 (UTC)
Having two "Spain" values is the way to go. Unfortunately, we don't have a way to attach references directly to sources. ChristianKl (✉) 20:21, 3 December 2017 (UTC)

According to the data model, the reference is for the whole claim, encompassing all qualifiers. If there is a disagreement like this one, two different claims should be made, albeit with the same value on the main snak, and then the respective reference given. --Denny (talk) 23:29, 3 December 2017 (UTC)

  • For Kuril Islands (Q80389), Japanese official position is that a.) all the islands involved in Kuril Islands dispute (Q389717), including Iturup (Q215727) and others, are not part of Kuril Islands (Q80389) (Source: [4], [5]), b.) For those islands that Japan recognized as part of the Kuril Islands, it is agreed that Japan have renounced rights on those islands in the 1951 treaty, however that does not mean Japanese government agreed that those islands should belong to Russia, instead their position is that future status of these islands should be determined by international conflict resolution mechanism in the future. (Source:[6](Japanese only)) C933103 (talk) 03:19, 4 December 2017 (UTC)

What to do regarding seemingly incorrect Item descriptions for a large number of listed building in the UK

Hi - I've been looking at items relating to listed buildings in the UK and it looks like a large number of them have incorrect descriptions. This one: Great Foldhay Farmhouse (Q26400424) is an example. It looks like the description field is actually the address for the building. I would have thought that a better description might be: "a Grade II listed farmhouse in Zeal Monachorum, United Kingdom". This would combine the properties of the item heritage designation (P1435) instance of (P31) located in the administrative territorial entity (P131) and country (P17) into a relevant and specific bit of text for the item. However, maybe these descriptions have been used for some specific reason? There are over 90,000 examples of these, so not trivial. Any advice welcome. JerryL2017 (talk) 08:14, 27 November 2017 (UTC)

You can use Descriptioner for mass description changes (SPARQL queries variables supported).--Jklamo (talk) 10:14, 27 November 2017 (UTC)
We could also delete most if not all descriptions and use automated descriptions in stead. We know for years that the descriptions are not good at all. Thanks, GerardM (talk) 12:42, 27 November 2017 (UTC)
Presumably they were used because they were convenient and they were better than nothing; there probably wasn't a description in the source they were imported from; and (@GerardM) there probably wasn't even a P31 statement for an automated description to start from. (Since then quite a lot of P31s have now been added in automated runs on the basis of the name; but there are still a lot of listed buildings with no P31, and a few for which the P31 that has been guessed may not be correct). Feel free to change the descriptions; but the address information is useful, so consider transferring that information to P969 (P969). Jheald (talk) 13:38, 28 November 2017 (UTC)
Thanks for all the input, I will look to update these descriptions. Regarding the suggestion around P969 (P969) from @Jheald it's a good idea, but I just wanted to say these addresses are only partial - this is typical: Barn And Outbuilding Approximately 30 Metres East Of Orchard Dene House (Q26300397) which has address: "Blewbury, Vale of White Horse, Oxfordshire, OX11". It could be combined with the label to give a more complete address, but in that case we end up with: "Barn And Outbuilding Approximately 30 Metres East Of Orchard Dene House, Blewbury, Vale of White Horse, Oxfordshire, OX11", which would be fairly typical and not really a bona fide address. Again, advice welcome. JerryL2017 (talk) 07:54, 30 November 2017 (UTC)

I have done lots and lots of work (tens of thousands of edits) trying to enhance this huge data import that was in many ways defective. Often, the P131 (located in the administrative division) was not right and pointed to an EU electoral district instead of a town, nearly always there was no P31 "instance of" (I added tens of thousands of that using the label), and the description is nearly never a description, but the mentioned address (which can't even be deleted after moving it to "located at address" since it's often necessary for dismbiguation). --Anvilaquarius (talk) 17:34, 4 December 2017 (UTC)

Guidelines for patronymic/matronymic systems of naming

Many countries follow patronymic (Q110874)/matronymic (Q1076664) systems of naming for persons, which is not the same as given(first) name/family name combination. family name (P734) currently has an alias 'last name'. However, this is not semantically correct since family name is not the same as patronymic/matronymic name. There must be a way to distinguish such names (may be with a qualifier). What could be the other ways? Personally, I do feel that we need to frame guidelines.

Also adding participants of Wikidata:WikiProject Names to the conversation.

  Notified participants of WikiProject Names. John Samuel 12:21, 2 December 2017 (UTC)

  • There are essentially two different schemes this concerns:
    • (A) The Icelandic (where this replaces the family name)
    • (B) The Russian (where this is added in addition to the given name and the family name.
(A) could fit in P734. As we haven't done much work on (B) yet, we might want to decide on that once other questions about these are sorted out.
--- Jura 12:27, 2 December 2017 (UTC)
  • Are you suggesting we have a new property Property:Patronym? I agree. We need it for Nordic/Scandinavian and Russian/Polish humans with names like Salmine Pedersdatter or Alex Abromovitz Dura. Do we also need a geographical identifier for older names that were before surnames were invented? Alex of Rhineland or Ole of Klugland. I am not sure of what that is called. In Scandinavia before surnames were used, you were identified by the farm you lived on or the city you were from. When you were forced to choose a surname, some chose the patronymic or chose their father's patronymic, and others chose the farm name. --RAN (talk) 18:37, 2 December 2017 (UTC)
There are two options: either a qualifier to a family name or new properties handling such occurrences of names. On checking Wikipedia entries of en:Patronymic and en:Matronymic, we can see that this practice is found in many countries of Africa, Asia and Europe. John Samuel 19:11, 2 December 2017 (UTC)
Is there a way to query which is (A) and which one (B)?
--- Jura 09:06, 3 December 2017 (UTC)
Does anyone know what the naming system is called for people like Isidore of Miletus (Q310028)? How would we identify all of these types of people in a search and what field should be added for these non-surnames?  – The preceding unsigned comment was added by Richard Arthur Norton (1958- ) (talk • contribs).
I think that there should be a Properties for patronymikon/matronymikon. Is there any reason strong for not proposing these two. Have they all ready been proposed? Breg Pmt (talk) 12:40, 3 December 2017 (UTC)
There's the open proposal https://www.wikidata.org/wiki/Wikidata:Property_proposal/Patronym_or_matronym where nobody offered any support or oppositon till now. It would be great if the proposal would get more input from people who understand the underlying issues and we can find a great label and description for it (at best a description that tells someone who doesn't know what a patronym is, what it is). ChristianKl (✉) 14:19, 3 December 2017 (UTC)
I suppose that we don't need patronymic property for Russians (and which: string or item??) - it is sufficient to have name in native language (P1559) and given name (P735) of father (P22). --Infovarius (talk) 16:57, 4 December 2017 (UTC)

Usage of the property "has parts of the class"

I have two questions about the usage of has part(s) of the class (P2670). When and what situation can we use this property?

  1. In my opinion, this is not the inverse relation of subclass of (P279), but sometimes used in this way (SPARQL). Is this valid?
  2. In order to describe the relationship between the instance of chemical compound (Q11173) (eg. water (Q283)) and chemical element (Q11344) (eg. hydrogen (Q556)), can we use has part(s) of the class (P2670), instead of has part(s) (P527)? I think has part(s) (P527) is inappropriate for this relation because, for example, not all hydrogen (Q556) are parts of water (Q283).

Thanks, --Okkn (talk) 16:21, 2 December 2017 (UTC)

I understand this property as being supposed to be used the way I did in set of intercostal nerves (Q1515961). It contains multiple intercostal nerve (Q27058076), so the property is appropriate. ChristianKl (✉) 16:28, 2 December 2017 (UTC)
The surgeon cut out a bunch of Intercostal nerves that it lays on his operating table. Is this "Set of Intercostal nerves"? --Fractaler (talk) 16:37, 2 December 2017 (UTC)
I consider this property to be about T1 to T11 whether or T12 / the subcostal nerve is included is more debateable and would need further specification (and maybe to separate items). ChristianKl (✉) 16:53, 2 December 2017 (UTC)
In the real world, surgeon or anatomist cannot isolate full set of and full length of intercostal nerves from the human body, so maybe that is a "part of set of intercostal nerves". --Okkn (talk) 16:59, 2 December 2017 (UTC)
The usage in set of intercostal nerves (Q1515961) seems appropriate for me. But, should has part(s) of the class (P2670) be always used in "set" item? Is there any other cases? --Okkn (talk) 16:45, 2 December 2017 (UTC)
Whenever an item has 200 meronyms that share a common superclass it doesn't make sense to list all the 200 with "has part" but using "has parts of the class" is better. ChristianKl (✉) 16:53, 2 December 2017 (UTC)
Sorry my English was unclear. I want to know the usage in the non-"set" item; for example, those in locomotive (Q93301) are appropriate? --Okkn (talk) 17:05, 2 December 2017 (UTC)
I don't think listing subclasses of the property as was done in locomotive (Q93301) is appropriate. A valid value for it might be something like "locomotive wheel" or other things appear multiple times inside a locomotive. ChristianKl (✉) 18:44, 2 December 2017 (UTC)
I'm in accord with you. What do you think about the relation between water (Q283) and hydrogen (Q556)? --Okkn (talk) 19:12, 2 December 2017 (UTC)
I think the water case is debateable. I think the current solution of "has part" works there. ChristianKl (✉) 20:10, 2 December 2017 (UTC)
water (Q283) now is chemical compound (Q11173). chemical compound (Q11173) has part(s) (P527) atom (Q9121). Conclusion: water (Q283) has part(s) (P527) atom (Q9121). hydrogen (Q556) is not hydrogen atom (Q6643508). --Fractaler (talk) 12:09, 4 December 2017 (UTC)
One way I like to use this property is when the subject has distinct components which are themselves classes. For example, three-piece suit (Q2998484) <has parts of the class> trousers, jacket, waistcoat. In this case the subject isn’t a set of like things, but rather a compound object with various sorts of parts. - PKM (talk) 19:40, 3 December 2017 (UTC)
Yes, three-piece suit (Q2998484) is a group of physical objects of three components: trousers (Q39908) (component 1), jacket (Q849964) (component 2), vest (Q958210) (component 3). Fractaler (talk) 12:09, 4 December 2017 (UTC)

Merging problem (restored)

Hello, I wanted to merge ordinance (Q2703920) to implementing regulation (Q3704604) as the entries within these items seem to cover more or less the same thing. However, there is one single conflict that causes blocking: Q3704604 with ja:行政委任立法 vs. Q2703920 with ja:命令 (法規). Does anybody have an idea how to resolve that conflict?--Herfrid (talk) 17:25, 23 November 2017 (UTC)

Merging them in ja.wikipedia. If these concepts are not "exactly" the same, probably they shoudn't be merged. strakhov (talk) 17:35, 23 November 2017 (UTC)
@Strakhov: Thank you for commenting. However, I'm afraid I'm not the right person to do that. On the other hand, it would be a shame if the merge would fail really just because of this single conflict, wouldn't it? I really don't know what to do…--Herfrid (talk) 17:42, 23 November 2017 (UTC)
You can do nothing. It often works. Somehow. :) strakhov (talk) 17:45, 23 November 2017 (UTC)
No, it's no shame if the conflict prevents you from merging two distinct concepts. Two concepts shouldn't be merged because they are more or less the same thing but only when they actually are the same thing. ChristianKl (✉) 17:49, 23 November 2017 (UTC)
But there only appears to be a problem with ja-Wiki – the other articles I managed to move to the apporopriate items. And I would still say that the entries within the two said items strongly match.--Herfrid (talk) 22:59, 23 November 2017 (UTC)
Let's see. Even if both articles in Japanese could be merged (I don't know that, I don't know Japanese), the items may need to remain independent and 'unmerged' (if they were not exactly the same), unless one of them is deemed as non notable (in that case it should be deleted). And that's not a problem for Wikidata. Our mission is far wider than providing interwiki links to wikipedias. strakhov (talk) 00:08, 24 November 2017 (UTC)
A strong overlap is not the same thing as two topics being the same. Whether or not the two things are the same is a question that needs expertise in international law to answer. I don't have that. If you have that expertise, feel free to make the argument but if you don't have the expertise to judge in a domain whether two things are the same, don't merge.
I outsourced the question to stackexchange, which is generally a good place to get expert opinions. ChristianKl (✉) 12:47, 24 November 2017 (UTC)
It looks like ja:命令_(法規) does mention the case in Japan, but ja:行政委任立法 does not, hence I   Oppose merging. --Liuxinyu970226 (talk) 10:04, 24 November 2017 (UTC)
@Liuxinyu970226: Thank you very much for forwarding the matter, though in this case, I actually do not think it is chiefly a question of labelling (as the commenter wrote, it had to read "ordinance" instead of "ordonnance", by the way), but whether the entries within these items really deal with different things or not, and as I can only assess by using Google translator this does not seem to be the case – not even with regard to the two Japanese articles, as ja:行政委任立法 is simply a one-to-one equivalent, i. e. a translation, of w:Statutory instrument, which simply focuses on the anglophone world (which could change, however, as soon as information about non-anglophone countries is included!). Hence, I still do not see a convincing rationale for having two separate items here.--Herfrid (talk) 15:02, 3 December 2017 (UTC)
The fact that jawiki has two articles is sufficient for not merging. If there's somebody with experience in law it might be interesting to actually contrast different notions of the legal concepts. To get more into the detail the German word "Verordnung" doesn't label things that are laws as far as I understand. According to enwiki statutory instrument is delegated legislation and delegated legislation is law. ChristianKl (✉) 19:39, 3 December 2017 (UTC)
Maybe @Takot: can tell us more about differents between two jawiki articles. --Liuxinyu970226 (talk) 10:34, 4 December 2017 (UTC)

Freedom of panorama

P3084 (P3084) is unused. Do we need it? Andy Mabbett (Pigsonthewing); Talk to Andy; Andy's edits 17:28, 3 December 2017 (UTC)

Seems quite hard to use: the property presumably would have some value if it could be used to tell whether a certain item can be photographed, but there are countless shades when you consider the kind of objects (defined in a different way in every country) and the way they are photographed (which again has some variance), plus the fact that multiple policies may apply (for instance a default rule for the country, specified by the entities). Deleting seems fine. Nemo 18:17, 3 December 2017 (UTC)
OK, see Wikidata:Properties for deletion#Property:P3084 Andy Mabbett (Pigsonthewing); Talk to Andy; Andy's edits 16:55, 4 December 2017 (UTC)

Duplicated zhwiki entry when linking articles to wikidata with zhwiki item

See https://i.imgur.com/EkFi8yg.png why there are two entry for "Chinese"? C933103 (talk) 02:40, 4 December 2017 (UTC)

Would you please link the item here? Matěj Suchánek (talk) 08:25, 4 December 2017 (UTC)
The "add links" widget was created back then it was only possible to link Wikipedia's. There might be more projects of the same language having the template, the widget only displays the language though. Sjoerd de Bruin (talk) 09:17, 4 December 2017 (UTC)
@C933103: Afaik, because both zhwiki and zhwikibooks are having Template:NoteTag, but now the problem is, why this interface can't mention something like "Wikipedia", "Wikibooks", etc. --Liuxinyu970226 (talk) 10:29, 4 December 2017 (UTC)

wikidata notability rule and link to the rule

I was trying to llink w:Template:NoteTag/doc on enwiki to w:ja:Template:Efn2/doc on jawiki using the sidebar tool, however I was warned with the text saying "Links to certain pages, like the one(s) you are trying to add (eg. template documentation subpages), are not allowed in Wikidata. Please check our rules. You can contact an administratorif you think you are correct."

  1. The links go to https://en.wikipedia.org/wiki/Special:MyLanguage/Wikidata:Notability , is it the intended behavior?
  2. Why Notability page does not exist despite implied as such by aforementioned link format?
  3. Why template documentations should not be linked via wikidata?

C933103 (talk) 02:50, 4 December 2017 (UTC)

Template documentation is not individually notable. It sticks with the parent page, to where it is transcluded.  — billinghurst sDrewth 03:52, 4 December 2017 (UTC)
Then should we use traditional interwiki links for cases that are different from this? C933103 (talk) 13:22, 4 December 2017 (UTC)
The warning is MediaWiki:Abusefilter-warning-subpages, which does link to WD:N. I tried to link a template subpage to an item and the link was correct. Not sure why the link to enwiki gets created. Matěj Suchánek (talk) 08:29, 4 December 2017 (UTC)
Maybe because the abusefilter error was shown on enwiki? Sjoerd de Bruin (talk) 09:15, 4 December 2017 (UTC)

Marking bot requests and requests for deletions as not done

I think the process of marking certain property proposals as not done is useful for reducing the amount of open request. Having a bot that waits a few days till the proposal is moved to the archive also allows people to easily object to the marking. Do we have a similar way to mark https://www.wikidata.org/wiki/Wikidata:Bot_requests or https://www.wikidata.org/wiki/Wikidata:Requests_for_deletions as "not done"? If not, should we have one? ChristianKl (✉) 14:19, 4 December 2017 (UTC)

On Wikidata:Requests for deletions you can use the template {{Not done}}, on Wikidata:Bot requests the template {{Section resolved}} to get archiving by bot. --Pasleim (talk) 14:54, 4 December 2017 (UTC)
Thanks. ChristianKl (✉) 15:27, 4 December 2017 (UTC)

Wiktionary mainspace sitelinks

Are these sitelinks allowed now, or is Wiktionary sitelinking just for project pages? Jc86035 (talk) 14:44, 4 December 2017 (UTC)

Shouldn't be active yet - you can see the video from the WikidataCon about what is planned and needed to get Wiktionary links here in a meaningful way [7]. Ahoerstemeier (talk) 15:42, 4 December 2017 (UTC)
No, this edit is wrong for several reasons. For starters, the IP added sitelinks to “nationalsocialism” on svwiktionary and “National Socialism” on enwiktionary, but while those terms describe the same concept, they’re not identical (they’re not the same lexeme), and so they shouldn’t be linked (svwiktionary could also have an entry for the English phrase “National Socialism”, and vice versa). They also shouldn’t be linked to Wikidata items (which are for objects and concepts), but to Lexemes (a separate entity type), and then there could be statements describing that one sense of the English lexeme “National Socialism” refers to the concept of nazism (Q7310). And finally, interwikilinks between Wiktionaries are managed by mw:Extension:Cognate, so as far as I understand, even once we have Lexeme entities on Wikidata, explicit sitelinks shouldn’t be added. So, yes, sitelinks should only be added for project pages (or templates, or probably some other kinds of pages that I’m forgetting right now), but not to the main namespace. --Lucas Werkmeister (WMDE) (talk) 16:11, 4 December 2017 (UTC)

Wikidata weekly summary #289

 
Here's your quick overview of what has been happening around Wikidata over the last week.
  • Development
    • Wikidata will get dedicated database resources, and go read-only for 30 minutes on 9th January 2018 (phabricator:T181645)
    • There were no RDF dumps last week due to problems generating them, investigation is still going on (phabricator:T181385)
    • Improved the threshold for ORES on Wikidata (phabricator:T180450)
    • Working on fixing a regression after a change in MediaWiki core that makes edit links show up on diff pages (phabricator:T181807)
    • More work on persistent editing of statements on Forms of a Lexeme (specifically phabricator:T180467)
    • Improved size of the diff that we sent to Wikipedia and co for changes happening on Wikidata. This is one more needed step towards only showing meaningful edits in the watchlists and recent changes there. (phabricator:T113468)

You can see all open tickets related to Wikidata here.


Decision: disable indexing of the userpages of Wikidata by robots

Tracked in Phabricator
Task T181525

Hello all,

After a message left on the Project Chat in French, I realized that currently, on Wikidata, the user pages are indexed for robots by default. Which means, their content can be found on search engines.

Usually, on the Wikimedia projects, this option is disabled by default. I created a ticket but first I'd like to ask your opinion about this. Do you think that we should change the existing parameter and disable the indexing of the user pages?

Thanks a lot for your answers, Lea Lacroix (WMDE) (talk) 11:38, 29 November 2017 (UTC)

  Support I did'nt even think that those could be indexed by robots, until the question was raised on the Bistro. I always thought that it was, like most other projects, not indexed by default. Please disable it. Also, is it the same for the Talk page, and the subpages ? --Hsarrazin (talk) 13:02, 29 November 2017 (UTC)
  Support Agreed to disable indexation of talk page by default for every user. Pamputt (talk) 15:24, 29 November 2017 (UTC)
  Support obviously, noindex should be the setting by default. Cdlt, VIGNERON (talk) 16:15, 29 November 2017 (UTC)
  Support I see no reason wikidata should be different. ArthurPSmith (talk) 16:58, 29 November 2017 (UTC)
  Strong support. Same as above. Nomen ad hoc (talk) 17:22, 29 November 2017 (UTC).
  Support YULdigitalpreservation (talk) 19:35, 29 November 2017 (UTC)
  Support Absolutely. - PKM (talk) 20:24, 29 November 2017 (UTC)
  Support Please. strakhov (talk) 20:33, 29 November 2017 (UTC)
"Usually, on the Wikimedia projects, this option is disabled by default." eh. that's not true. That's just something that a few of the bigger wiki's have decided because they have spam problems which they don't want to effect their reputation. TheDJ (talk) 21:12, 29 November 2017 (UTC)
I personally think this is 'breaking the internet' and eventually pointless. If you have stuff you don't want index'ed, you should remove it. TheDJ (talk) 21:21, 29 November 2017 (UTC)
  Support, no reason to index user pages--Ymblanter (talk) 21:13, 29 November 2017 (UTC)
  Oppose the statement "Usually, on the Wikimedia projects, this option is disabled by default" seems to be incorrect. Where did you base this on Lea? This seems to have misled a lot of people in supporting this. I actually use search engines to find stuff on user pages because our own search engine isn't that good. Multichill (talk) 21:14, 29 November 2017 (UTC)
Let's amend that: We seem to have set quite a few norobots for user pages on different projects, but these get indexed anyway. So I'll just change this to {{I-don't-care-because-it-doesn't-work-anyway-so-let's-just-do-it-because-it-doesn't-break-anything}}. Multichill (talk) 21:22, 29 November 2017 (UTC)
Actually, looking at the 'partial' Mediawiki:robots.txt is not enough to check that. You are looking for 'NamespaceRobotPolicies' of the configuration. TheDJ (talk) 21:20, 29 November 2017 (UTC)
Thanks for the links and more precise information. Indeed, it is less widespread than I thought. I'm not sure that this really influences people to decide what's happening on Wikidata anyway. Lea Lacroix (WMDE) (talk) 09:01, 30 November 2017 (UTC)
  Support Good idea.--Jklamo (talk) 09:27, 30 November 2017 (UTC)
  Support --β16 - (talk) 09:37, 30 November 2017 (UTC)
  Support I always wonder why this is not Wikimedia's default. Matěj Suchánek (talk) 16:08, 30 November 2017 (UTC)
  Support Good move. Lymantria (talk) 06:22, 1 December 2017 (UTC)

Is it meaningless to put copyright symbols on datasets?

Hi all

I've been putting together some resources on Wikidata around data rights information for data producers, my understanding is that data does not fall under copyright and have multiple reliable sources that say this. However lots and lots of organisations put copyright notices on their data, does this simply come from a lack of understanding of database rights?

Thanks

John Cummings (talk) 20:31, 4 December 2017 (UTC)

I believe the general rule is that individual items of data are not copyrightable, but a collection of data can be copyrighted. One phone number cannot be copyrighted, but the phonebook can be, etc. ArthurPSmith (talk) 21:25, 4 December 2017 (UTC)
The United States would beg to disagree, see Feist v. Rural (Q5441583). In fact, the EU wouldn't find copyright in a database either, without some subjective choices having been made as to or in its data. But in the EU the collection might be covered by database rights. Jheald (talk) 21:08, 5 December 2017 (UTC)
Disclaimer: IANAL. To my layperson knowledge the protection of a work depends mostly on the applicable law(s) in the jurisdiction the creator/holder of rights is based in, not on the formal use of copyright symbols. This might have been different in the past (which may be the reason why people are used to use those symbols), and some subtle consequences may still result from the use of copyright symbols. This is not a specific situation for data/data sets/databases, but rather valid for all kinds of works. —MisterSynergy (talk) 21:26, 4 December 2017 (UTC)

Update: After some more research it appears that some national copyright laws cover databases when they include creativity, I've updated the page to reflect this. John Cummings (talk) 22:22, 4 December 2017 (UTC)

How to Use Username

Hello, I am trying to create an account on Wikidata. I searched there is no person found named "Nikhil Bansal", but when I enter the name Nikhil Bansal, its show name already in use. Please let me know what is the policy to choose the username.

Thanks  – The preceding unsigned comment was added by 112.196.43.21 (talk • contribs) at 05:49, 5 December 2017‎ (UTC).

Such an account was indeed registered at Metawiki in May 2016, but it never visited Wikidata or made any edits. Account names have to be unique across all Wikimedia projects, so this one would no longer work. We do not have a specific Wikidata username policy, but most aspects of the English Wikipedia version would apply here as well. —MisterSynergy (talk) 07:05, 5 December 2017 (UTC)
You will need to keep trying username alternatives until you find one that is available, if you want to research availability try Special:GlobalUsers.  — billinghurst sDrewth 05:20, 6 December 2017 (UTC)

Search misses top item

When I use the top search box to look for "Gold", the result list does not mention the most obvious item (gold (Q897)). It does list many items with "gold" in a context (top was "Gold (Gold (Q13553585)): family name"). How come? -DePiep (talk) 11:34, 5 December 2017 (UTC)

The search function never worked as it is supposed to. Many a user created a duplicate item because the search did not turn up the main item. - Brya (talk) 12:01, 5 December 2017 (UTC)
When I search it does show gold (Q897) on the top. My language version is set to English, what language do you use for Wikidata? If you set it to Italian it does show Gold (Q13553585) first because it has the name Gold in Italian while gold (Q897) doesn't. ChristianKl (✉) 12:10, 5 December 2017 (UTC)
With English as selected language, I see gold (Q897) as the first element in the dynamic dropdown list of the search box; however, it does not appear at all on the search page within the first 20 results. —MisterSynergy (talk) 12:58, 5 December 2017 (UTC)
Searching for English terms when using a non-English user interface has become quite some issue. Exact matches are lower in the list, while scientific publications that start with the term are above them. @Smalyshev (WMF): can we boost English labels more, or decrease those publications? Sjoerd de Bruin (talk) 15:13, 5 December 2017 (UTC)
@Sjoerddebruin: It is possible technically have boost to English language matches (though I am not 100% happy doing special-case rankings, but maybe in this case it's warranted). Note however for items like Gold (Q13553585) or 4955 Gold (Q1044365) "gold" would match local language - e.g. it has Russian alias "Gold" - so the match would be legit. I don't think we want to rank English match higher than Russian match in this case, but maybe we could do other tweaks. We'll look into it. Smalyshev (WMF) (talk) 18:49, 5 December 2017 (UTC)
This is a tricky question. First of all, there are two searches - the prefix search (autocompleting one in the top box) and fulltext search (Special:Search). The latter one right now doesn't work very well, we're working on fixing it (see phab:T178851) but it'll take some time. The former one should work better, but there is a dilemma - what is better - the match in the language you asked for (i.e. label in Italian if you search in Italian) or label in different language (i.e. English) but with high popularity score (a lot of links, etc.)? Right now in this case it looks like language match is winning - i.e. if the label in Italian is "gold" (including aliases) then it shows up first. And that's the case for Gold (Q13553585). We could tweak the scoring but it could make other searches perform worse.
I'd also want to know what is the reason to search for English name for gold (Q897) while having interface in non-English? Is it some special use case or just because more items have names in English? Smalyshev (WMF) (talk) 18:43, 5 December 2017 (UTC)
At some point I forgot the Dutch translation of some items... Sjoerd de Bruin (talk) 18:50, 5 December 2017 (UTC)

Quora questions polluting Mix-n-Match

Crazy stuff like https://www.quora.com/topic/When-a-Balloon-Is-Rubbed-with-a-Sweater-Which-Charge-Does-the-Balloon-Has does not belong to and is polluting Mix-n-Match (and some of these questions are already deleted on Quora!).

Please comment here how we can distinguish Quora topics from questions, or whether it's also ok to delete topics from MnM (of course, Quora topic ID (P3417) instances will stay).

Copying @Pigsonthewing: who proposed this property. --Vladimir Alexiev (talk) 16:02, 5 December 2017 (UTC)

Maybe we could delete any unmatched item with more than three '-' from the list for Quora? ChristianKl (✉) 16:29, 5 December 2017 (UTC)
Please respond on the given thread. Sjoerd de Bruin (talk) 16:31, 5 December 2017 (UTC)

Identifier links and nofollow

Tracked in Phabricator
Task T175230

As explained at mw:Manual:Nofollow, for many years MediaWiki on Wikimedia wikis has added the "nofollow" attribute to external links (except some whitelisted targets) in the hope to reduce the incentive for link spammers. Some months ago I noticed the external links are not treated consistently in Wikidata, namely the external links from the identifiers section are treated like internal links: phabricator:T175230. I've not had any comment yet.

I suspect some users may not be aware that, unlike adding a normal external link, an identifier donates a dofollow link (i.e. presumably pagerank) to the target website, which can e.g. be a for-profit. I'm not sure it's ok for a non-profit commons like Wikidata to donate valuables to for-profits. I also wonder whether this configuration mistake has already been exploited by some spammers who noticed it, or has otherwise encouraged spam on this wiki. --Nemo 12:58, 5 November 2017 (UTC)

  • It might be a reason why we get so many marginally useful "do-it-yourself" identifiers.
    --- Jura 13:19, 5 November 2017 (UTC)
  • I don't think there's a problem in principle with setting dofollow links and I don't think it should be thought of as a donation. It's a variable that Google can use to know that a given resource is a valuable reference. When it comes to external-ids we make a decision whether a given link is welcome in our project when creating the property. If we create a external-id property for www.TheBestReferenceEver.com/$1 and there's no problem with creating an incentive for the owners of TheBestReferenceEver to do the work of creating a bot that enters statements for relevant items. Such a bot isn't spamming but providing a valuable service.
    From a SEO perspective I however think it dilutes the values of the links we do set to internal Wikimedia project and thus it might be reasonable to set "formatter urls" as non-follow. ChristianKl (talk) 14:18, 5 November 2017 (UTC)
    • All identifier links become a problem if they don't have a nofollow while all the other links do, but especially so in the case of for-profit and non-free websites, because their content is worthless from a free knowledge and linked open data point of view: the mere fact that an identifier exists and corresponds to that item is information, which can be hosted on Wikidata so as to reduce duplicated work out there and to allow someone (e.g. some Wikipedia entries) to use it; but following the URL itself does not add any free information to the item. --Nemo 19:26, 3 December 2017 (UTC)
  • It's also worth noting that while EnWikipedia pages have high pagerank and thus dofollow links matter a significant amount the same isn't true for Wikidata pages. ChristianKl (talk) 15:05, 5 November 2017 (UTC)
  • Excessive and gratuitous use of nofollow is contray to the spirint of the web and an evil to be avoided. Andy Mabbett (Pigsonthewing); Talk to Andy; Andy's edits 11:07, 6 November 2017 (UTC)
  • +1 instead of passive-aggressive methods to "discourage" external links, maybe some link curation would add to quality. but clearly there is at least one admin who would rather edit war with every ip, and filter every librarian, than engage in collaboration. time to rethink and reconsider. Slowking4 (talk) 04:18, 7 November 2017 (UTC)
    • I agree, but how would you apply "link curation" to the case of identifiers? Once an identifier property is approved, it's not like you can prevent people from using it. An identifier link is goes towards comprehensiveness, not relevance: if we strive for link quality, therefore, identifier links are worth a nofollow even more than the other links, which are supposed to be curated as you say. --Nemo 13:50, 11 November 2017 (UTC)
      • Could you give examples where you feel there's a problem with identifier links, where you agree with an identifier link existing but disagree with it existing in a particular case? ChristianKl (✉) 14:16, 11 November 2017 (UTC)
        • My example would be "100 % of cases", see above. But I'll try to answer anyway: on a page like Q13879663, it's ok to know that a subreddit was created for it, but the subreddit itself is empty; on Q27928396 it's good to know about the two hashtags, somebody might even want to link them; but it's nonsense that Twitter is in fact the primary URL for the page (the only URL with dofollow) while the most interesting URL to follow is clearly the official website (not to mention that [8] right now also contains anti-immigration propaganda, ads and mere spam. --Nemo 19:26, 3 December 2017 (UTC)
          • If you say you do agree with the fact that the Twitter hashtag does exist, what kind of standards would you like to see to decide when it's used? I think that's a question we have to discuss directly instead of focusing on the indirect way of playing with incentives. ChristianKl (✉) 11:52, 5 December 2017 (UTC)
            That's my point, you cannot decide when it's used. The information about the hashtag is correct, just the link is useless. When you mix the two problems, they become impossible to solve.
            If we could decide to linkify identifiers on a case by case basis, then I'd say that users should make an identifier a link only if there is no other URL on the item, or when the URL will be the best representative URL for the subject of the item; and only after checking that there is relevant, good content at the URL, with no expectation of spam coming up. Of course this is not practical; we must just acknowledge that those links generally have no value. Nemo 07:52, 6 December 2017 (UTC)
              • If you advocate that the link generally has no value the obvious step would be calling for the removal of the links directly. That could be easily done by removing the formatter url. But that's not what you advocate when you focus on dofollow. ChristianKl (✉) 18:15, 6 December 2017 (UTC)

dyad etc.

The item dyad (Q29431432) is spread by User:Fractaler at items that consist of two objects. Example: biological sex (Q290) is subclass of dyad. I think this case is fatal, because there are more than two sexes. Also residence (Q699405), redox (Q82682) and many more are affected.

Further dyad is subclass of abstract entity (Q7184903), which means that e.g. house (Q3947), bicameral legislature (Q189445) is also subclass hence abstract.

I had a discussion about matrices, but we didn't find a agreement.

My opinion: stop spreading dyad etc. and filter these items by query "consists of two objects". But I think the current situation is untenable.

@Fractaler that is not meant personally I think you are doing a great job. Greetings Bigbossfarin (talk) 15:07, 1 December 2017 (UTC)

I don't see a good reason to have biological sex (Q290) subclass dyad (Q29431432). For items such as pair of human eyes (Q41890371) which are actually about a group of two objects I find it okay. Having Seven Wonders of the Ancient World (Q489772) subclass heptad is also fine with me. ChristianKl (✉) 17:38, 1 December 2017 (UTC)
those don't sound like subclass but instance relationships to me ArthurPSmith (talk) 19:47, 1 December 2017 (UTC)
@ArthurPSmith: same to me. Changing this would also solve the problem that a house (Q3947) is instance of an abstract object. Bigbossfarin (talk) 20:10, 1 December 2017 (UTC)
Bigbossfarin, wrong, items don't consist of objects. Must I say what consist of objects, items? Next. biological sex (Q290): "specialization of living organisms into a male or female variety". 1+1=2? Next. dwelling= Q7743 + lodging. 1+1=2? Next. redox (Q82682)=Q1786087+Q912546. 1+1=2? Are you frustrated that the group is an abstract object? Then maybe you can suggest a better option so that a group of objects is not an abstract object. How about non-abstract object. And why have not you answered the questions that I asked you on the matrix page? --Fractaler (talk) 20:20, 1 December 2017 (UTC)
You can describe a dwelling in many different ways of consisting of any different number of parts that you. It's qualitatively different from an item like pair of human eyes (Q41890371) which has no clear parts and Seven Wonders of the Ancient World (Q489772) which has clearly 7 parts. ChristianKl (✉) 03:26, 2 December 2017 (UTC)
Do you mean that "dwelling" is homonym (Q160843) (first, when 1+1; second, when 1+1+1; third, when 1+1+1+1; etc.)? So I always said: for today in Wikidata there are full of homonyms, because of which the Wikidata are in chaos. And the problem of homonymy is solved by the usual division of the homonym into non-homonyms. --Fractaler (talk) 15:27, 2 December 2017 (UTC)
No, I don't mean that it's a homonym. Let's say you have X with contains A, B and C. Now you come and say it's a dyad because it contains (A and B) and C. In addition even if it would be a homonym, division into homonyms has a cost. It increases complexity because now whenever adds a new item with instance of (P31) dwelling, they have to decide which of the two they should add and whether they should add both. If you get your way they might have to add 10.
The added complexity that makes it harder for new users (and also existing ones) to add new dwellings isn't worth the little benefit that the structure that you propose might add. ChristianKl (✉) 18:58, 2 December 2017 (UTC)
X with contains A, B and C: "consists of" (components) = "contains"? An iron bucket can contain water, but not consist of water. An ice bucket may contain liquid water, but not consist of liquid water. But an ice bucket with liquid water (as a single object, whole, a system) consists of water molecules. --Fractaler (talk) 06:38, 4 December 2017 (UTC)
Of course Bigbossfarin is right. dyad (Q29431432) now is homonym (Q160843), because is "group or series of two (real, mental, virtual, etc.) objects". How to solve the problem of homonymy is known - clarification. Ie, in order for a physical object to fall only into the physical (and not mental, mathematical or virtual) space, it is necessary to specify the terms - physical dyad, mental dyad, virtual dyad, etc. This, of course, is an obvious fact, a logical thing that brings order to the Wikidata. But I consider it my duty to warn - such obvious things here are deleted without discussion (order in the wikidata is not yet required, the structuring of data is not yet the goal of the Wikidata). --Fractaler (talk) 15:27, 2 December 2017 (UTC)

@Fractaler, ArthurPSmith, ChristianKl:
Let us vote to find a compromise. Bigbossfarin (talk) 13:46, 3 December 2017 (UTC)

Who is pro/contra changing subclass to instance dyad (Q29431432)?
  Support --Bigbossfarin (talk) 13:46, 3 December 2017 (UTC)
  Support though there may be exceptions that make sense to retain as subclass relations ArthurPSmith (talk) 15:47, 4 December 2017 (UTC)
Where should it be used?
item pro contra
biological sex (Q290) Bigbossfarin ChristianKl ArthurPSmith
residence (Q699405) Bigbossfarin ChristianKl ArthurPSmith
testicle (Q9384) Bigbossfarin ChristianKl
complex number (Q11567) Bigbossfarin ArthurPSmith
Euclidean vector (Q44528) Bigbossfarin ArthurPSmith
yin and yang (Q62744) Bigbossfarin ChristianKl ArthurPSmith
redox (Q82682) Bigbossfarin ArthurPSmith ChristianKl
biathlon (Q166788) Bigbossfarin ArthurPSmith
bipartite graph (Q174733) like any graph - V+E. Infovarius Bigbossfarin ChristianKl
binomial (Q193623) Bigbossfarin ArthurPSmith
bicameral legislature (Q189445) Bigbossfarin ArthurPSmith
Nordic combined (Q201965) Bigbossfarin ArthurPSmith
two-wheeler (Q233040) Bigbossfarin ChristianKl (A two wheeler has more parts than just the wheels)
two-state quantum system (Q243821) Bigbossfarin ArthurPSmith, Infovarius
double bond (Q258652) Bigbossfarin ArthurPSmith
diol (Q421916) Bigbossfarin ArthurPSmith
dimer (Q531306) Bigbossfarin ArthurPSmith
left and right (Q542952) Bigbossfarin ArthurPSmith
wheel and axle (Q592289) Bigbossfarin ChristianKl ArthurPSmith
Cooper pair (Q620057) Bigbossfarin ArthurPSmith
lone pair (Q804644) Bigbossfarin ArthurPSmith
I have not voted on cases where I was uncertain. I voted opposed in most of the quantum-mechanics related cases as I think the quantum concept of identity and pairing etc. is a very different thing conceptually from the "dyad" of something made up of two independent pieces. Also I'm not really comfortable with using the word "dyad" or applying these relationships at all here, it has far too many meanings - see en:Dyad for instance. ArthurPSmith (talk) 15:47, 4 December 2017 (UTC)

Scientific approach

item Model 1.1 Model 2
Biological universe/space/world
male or female (biological role/function) biological monadic role/function -> biological role/function. Eg, "boy (Q3010)" does not have the role/function "male (biological role/function)" your model
male or female (biological feature) biological monadic feature -> biological feature. Eg, "boy (Q3010)" has the feature "male (biological feature)" and does not have the "male and female (biological feature)" your model
male and female (biological feature) biological dyadic feature -> biological feature. Eg, hermaphroditic organism has the dyadic feature "male and female (biological feature)" your model
testes (two testicles) paired/dyadic reproductive gland -> dyadic component of reproductive system -> dyadic component of organ system -> dyadic component of biological system -> dyadic component of physical system -> component of physical system your model
Mathematical universe/space/world
complex number (Q11567) number consisting of 2 components -> number consisting of components and mathematical object consisting of 2 components -> mathematical object consisting of components -> mathematical object your model
bipartite graph (Q174733) (dyadic graph) mathematical object consisting of 2 components -> mathematical object your model
physical dyadic object: "physical object consisting of 2 components", PDO
dwelling=lodging+

home (building))

PDO -> physical object consisting of components -> physical object your model
dimer (Q531306) (dyadic oligomer) dyadic molecule -> physical object, consisting of identical components -> physical object consisting of components -> physical object your model
.

In science there is no compromise, authority (temporal consensus, source of approval, etc.). Voting is an unscientific approach. In science, if the model is bad, then attempts are made to suggest a better option. If you have a better model, you simply replace the words "your model" with yours model. --Fractaler (talk) 11:18, 4 December 2017 (UTC)

Wikidata is at it's core not about doing scientific original research but a secondary source. Part of creating a secondary databases is that authorities do matter. Statements that aren't supported can be deleted without any need for adding other statements. ChristianKl (✉) 12:03, 4 December 2017 (UTC)
Ie, the Wikidata do not belong to science, and therefore, belong to the non-science, right? Fractaler (talk)
If you take a science textbook, that's not non-science but at the same time the textbook also doesn't contain original research but findings that have been proven by findings that are published in authoritative journals. In general providing good references is an essential part of how our scientific system works when it communicates knowledge. ChristianKl (✉) 12:55, 4 December 2017 (UTC)
Ok, what is 1) "non-science"; 2) "science" by your version? Fractaler (talk) 12:59, 4 December 2017 (UTC)
I don't feel the need to define them. I do like Kuhn's notion of it. Instead of speaking about what's science and what isn't it makes more sense to speak about what kind of policies Wikidata persues and Wikidata intents to focus on being a secondary source and not a primary source that publishes original research. ChristianKl (✉) 14:55, 4 December 2017 (UTC)
Does the user need references to secondary sources to detect the Wikidata statement that, for example, dimer (Q531306) is an abstract object (through the Wikidata verification tool: dimer (Q531306) -> dyad (Q29431432) -> group (Q16887380) -> class (Q5127848) -> abstract entity (Q7184903))? --Fractaler (talk) 07:19, 5 December 2017 (UTC)
"Does the user need references" is the wrong question. A better question would be whether other sources actually see the item as a dyad (Q29431432). The act of seeking sources and synthesizing knowledge from multiple sources is quite helpful. If you would go and research you would likely find it hard to find sources that say that "dwelling" is a physical object with numeric parameter = 2. If you would ask most people on the street whether "dwelling" is a physical object with numeric parameter = 2 or 3 they wouldn't know what you are talking about.
Saying that something is a "dyad" includes saying that it's a group. Most people don't consider a graph to be a group and as a result I don't think it makes sense to label bipartite graph (Q174733) a group. If you would find people who actually think of a graph as a group it would be more interesting and we could discuss what kind of assumptions they make to come to their conclusion. ChristianKl (✉) 18:57, 5 December 2017 (UTC)

I think it's a horrible idea to use instance of (P31) or subclass of (P279) here. At most, belonging to a group of two is a characteristic of an object. It is not the object's defining essence. P31 and P279 are not appropriate. Jheald (talk) 21:16, 5 December 2017 (UTC)

@Jheald: So we can count you as voting "Against" in all the items on the list? ChristianKl (✉) 12:59, 6 December 2017 (UTC)
totally agree with @Jheald: on this.   Oppose to all... --Hsarrazin (talk) 13:17, 6 December 2017 (UTC)

Wikidata deployments will now happen every week

Hello all,

Some information that may interest people who contribute to Wikidata and Wikibase, or follow the development.

We made some changes in our deployment structure, the most relevant things for you are:

  • we are now able to deploy new code for Wikidata every week, instead of every two weeks. test.wikidata.org will be updated on Tuesdays and wikidata.org on Wednesdays (details)
  • beta.wikidata.org is now updated every 10min with new code (details)

Which basically means that new features and bug fixes will arrive faster to you :)

If you have questions or want further details, feel free to ping me. Lea Lacroix (WMDE) (talk) 14:55, 6 December 2017 (UTC)

Wikidata (Q2013) is not a data store, Wikimedia Commons (Q565) is a data store?

For example, Data 1 Data 2, Data 3 (Wikidata has only a link list of U.S. states and territories by GDP (Q3241935)). Right? --Fractaler (talk) 08:20, 1 December 2017 (UTC)

Which sources do you have as a basis for your claim? ChristianKl (✉) 09:45, 1 December 2017 (UTC)
Why "claim"? "?" = just "question". --Fractaler (talk) 10:04, 1 December 2017 (UTC)
The answer to the question is "whatever you can show with sources". ChristianKl (✉) 17:39, 1 December 2017 (UTC)
Sorry, I do not understand a question. Maybe I'm not clearly speaking. then this question: what are Data 1,Data 2, Data 3, etc? Data? --Fractaler (talk) 14:50, 2 December 2017 (UTC)
“Data” is a very generic term which does not tell you anything about Wikidata or Commons. “Data” does not even have to be digital data. So the best method to manage “data” strongly depends on its type.
In the Wikimedia universe, “Pixel data” for instance clearly fits to Commons and unstructured text to Wikipedia. Structured data goes to Wikidata, while bulk tabular data (and most raw data) does not really fit in here. You can of course try to squeeze it into Wikidata somehow, but it will be a pain to reuse it and thus of little value. Tabular data such as the NYC weather history is therefore better stored at Wikimedia Commons.
So in fact both Wikidata and Wikimedia Commons are data stores. They both have a certain focus and unique features and cannot replace or outperform each other. —MisterSynergy (talk) 17:05, 2 December 2017 (UTC)
Ordered list (for example, this) is a structured data? --Fractaler (talk) 13:56, 4 December 2017 (UTC)
Where should the list be stored (here or on Wikimedia Commons (Q565))?? --Fractaler (talk) 07:52, 5 December 2017 (UTC)
The list table, as it is, should stay at Commons. (It can’t be uploaded here in its current format anyway.) Nevertheless you can add individual information from the list table at Wikidata as well. Users how want to compare GDP by state will use the tabular data from Commons, while users interested in a particular value for a given state will look for the claim here at Wikidata. —MisterSynergy (talk) 10:06, 5 December 2017 (UTC)
What is "list, as it is"? Q18616576#P1963, for example, is "list, as it is"? --Fractaler (talk) 10:20, 5 December 2017 (UTC)
Sorry, wrong word; I meant "table". —MisterSynergy (talk) 10:24, 5 December 2017 (UTC)
Ok, 2D data (vertical list x horizontal list=table, table of numbers, symbols, pixels, etc.) should stay at Commons. How about 1D data (list)? --Fractaler (talk) 10:38, 5 December 2017 (UTC)
You can use the “dimension” of a data set (in tabular for) as an indicator, but not as the only one. Other ones to consider would be “bulkiness” and “density” of the data set (also in tabular form):
  • Densely filled tables shall stay tabular at Commons, sparsely filled ones might better fit to Wikidata.
  • Bulky tables of any dimension (with many rows/columns in at least one dimension) should stay tabular at Commons if they are not sparsely filled.
  • There are no fixed boundary values which definitely tell you which way to go; both ways, tabular data at Commons and structured data at Wikidata, have advantages and disadvantages, so and it depends on the use case which one to use (or where to upload).
  • Mind that tabular data at Commons and structured data at Wikidata are by no means mutually exclusive. There may be use for both formats, and it has to be decided for many (text-based) data sets individually what to do. A scenario would be to have bulk tabular data at Commons, and to add some selected values out of it as claims in a Wikidata item as well.
So to answer your question: a lengthy list (1D table) can also stay at Commons if this is more useful than Wikidata statements. —MisterSynergy (talk) 11:36, 5 December 2017 (UTC)
List, tabular data at Commons are not a structured data? Then what is a structured data? --Fractaler (talk) 11:48, 5 December 2017 (UTC)
Although it has of course some structure, tabular data at Commons does not have an RDF data model, unlike “structured data” at Wikidata has. —MisterSynergy (talk) 13:25, 5 December 2017 (UTC)
So, "structured data" = "data with an RDF data model"? Fractaler (talk) 14:21, 5 December 2017 (UTC)
In this topic: yes. Please be aware that data with less structure also sometimes referred to as “structured data”, although an RDF data model is missing. —MisterSynergy (talk) 14:35, 5 December 2017 (UTC)
"tabular data at Commons does not have an RDF data model, unlike “structured data” at Wikidata has". So, tabular data at Commons is 1) “structured data” or 2) “unstructured data”? Fractaler (talk) 06:30, 6 December 2017 (UTC)
Again the question isn't usefully framed. Black and white thinking isn't useful here. John Cummings wrote a nice summary: https://www.wikidata.org/wiki/User:John_Cummings/Publishing_open_data the data inside Commons has 3 stars in that framework while Wikidata has 5 stars. ChristianKl (✉) 13:53, 6 December 2017 (UTC)
Authorities (stars, references, etc.) need religion and pseudoscience. Science does not need authority. And why 1 or 0? I'm just asking: "tabular data at Commons is “structured data”. There are 4 answers: 1) yes, 2) no, 3) I do not know, 4) "structured data" is homonym (Q160843). --Fractaler (talk) 14:08, 6 December 2017 (UTC)
Your scope is far too narrow, your stupid insistence on black vs. white, while ignoring the grey and colorful reality, makes any conversation with you futile. First you have to define "structured", as there are many ways to define this. Structured for the human eye, structured for dumb machines, structured for the human ear, structured in space, time ort spacetime? Strict or vague structures, i.e. with clear borders or blurred border areas? Your answer is plain wrong, as it looks for an answer that simply doesn't exist. Sänger (talk) 14:31, 6 December 2017 (UTC)
makes any conversation with you futile: so why are you joining such a conversation? And your 285-character "First ...areas?" = 1-character answer "4". --Fractaler (talk) 06:34, 7 December 2017 (UTC)

Privacy issues in our policies

I'm at the moment writing on a policy for Living people and will soon open an RfC for approving the policy. I will also address at least one other privacy issue in the RfC given that it's a good time to raise them. If you can think of privacy issues, that are currently unsolved in our project and not addressed by the existing draft, I'd appreciate if you can alert me of the issues that you see (even if I might disagree ;)). ChristianKl (✉) 20:59, 5 December 2017 (UTC)

@ChristianKl: I am of the opinion that sex or gender (P21) must be included under the list property likely to be challenged (Q44597997). Some people may prefer not to disclose sex or gender (P21) and may consider this as sensitive personal information. Currently, I do not see this being considered and often seen as a mandatory information. John Samuel 18:23, 6 December 2017 (UTC)
Thanks, I think that's a discussion that is to be had on the discussion page of sex or gender (P21) after the RfC is concluded. I want to focus the RfC more on having a general process for adding and removing items for the list for property likely to be challenged (Q44597997) than it being about the the exact contents of the list. ChristianKl (✉) 19:25, 6 December 2017 (UTC)

How do RfCs get archived?

Given our huge RfC backlog, I closed a few. There seem to be multiple possible ways to close them. Do we have bots that archive closed RfCs? If so, how long does the bot wait? What information does the bot need to be triggered? Which bot does the job? ChristianKl (✉) 14:23, 6 December 2017 (UTC)

I don't think there's a bot, you have to add their resolution to the table on the main RfC page. ArthurPSmith (talk) 16:42, 6 December 2017 (UTC)
On EnWiki there's a bot and I think that makes generally sense. I'll start here by waiting a month and see whether any bot does something. If there's really no bot that does the talk I would advocate to copy the way EnWiki does it. ChristianKl (✉) 20:47, 6 December 2017 (UTC)

Fusionproblems

Can someone fusion en:Sierra Leonean general election, 2018 (Q26709117) with de:Parlamentswahl in Sierra Leone 2018 ? --178.3.20.159 17:40, 6 December 2017 (UTC)

Properties for location maps modules / templates

I would like to ask for your thoughts about items which are Wikipedia location maps' modules/templates (such as Warsaw location map template (Q6714634)). I think it would be useful if we could add some properties to those items, especially a property which would say what country / administrative unit / city etc. is actually on the map. I am willing to spend some time on this, but my question is: which property would be most suitable for that? And are there any other properties that you would add to such items? Thank you in advance for your opinion. Powerek38 (talk) 17:48, 6 December 2017 (UTC)

Member of a legislature or council

Could I ask someone to check whether what I did at Jan Drago (Q6148794) with position held (P39) is the correct way to handle this, or if it should be done differently? - Jmabel (talk) 07:29, 13 December 2017 (UTC)

This looks good to me. Andrew Gray (talk) 12:15, 13 December 2017 (UTC)
This section was archived on a request by: Jmabel (talk) 16:10, 13 December 2017 (UTC)

Reconciling with Wikidata via third-party identifiers

I have a set of identifiers to add to Wikidata, and attached to these records are links to third-party sources of information about each item. These might be Wikipedia pages in English or German, might be Oxford Dictionary of National Biography ID (P1415), GND ID (P227), CERL Thesaurus ID (P1871), VIAF ID (P214) or similar. I am already a user of QuickStatements and Wikidata tools for Google Sheets. With Google Sheets I can generate Q numbers from the Wikipedia links. Where I need help is with an efficient way to match using the third-party identifiers. It's not practical to download all of Wikidata's VIAF ID (P214) relations and do a lookup. I was thinking of using the query service; construct a URL to the query service, return some XML or JSON to the Google Sheet, then strip out everything but the Q number. Is there a simpler tool or process that I'm missing? Thanks in advance for any help, MartinPoulter (talk) 15:57, 13 December 2017 (UTC)

You might want to look into OpenRefine - the latest release has wikidata reconciliation built in where you can specify a property (like VIAF) associated with a column in your data and it auto-matches to the right wikidata item. WikidataIntegrator (python bot library) also does something similar if you want to do some python coding. ArthurPSmith (talk) 15:59, 13 December 2017 (UTC)
See Wikidata:Tools/OpenRefine for details. ArthurPSmith (talk) 16:12, 13 December 2017 (UTC)
Sounds just what I want. Thanks, MartinPoulter (talk) 16:35, 13 December 2017 (UTC)
This section was archived on a request by: MartinPoulter (talk) 16:53, 13 December 2017 (UTC)

Vandalism

Category:Abbreviations

Aktogay District (disambiguation page)

subclassOf smartphone (remove P279, leaving items outside the P279-classification tree)

92.226.213.195 14:08, 7 December 2017 (UTC)

More, located on terrain feature

85.182.112.199 15:23, 7 December 2017 (UTC)

I have to say I agree that it's perfectly reasonable for each mobile phone model to be subclass of (P279) mobile phone (Q17517) or an appropriate subclass like smartphone (Q22645). We have similar relationships with automobiles for example. The P31 relationship should be with the metaclass "mobile phone model". ArthurPSmith (talk) 15:40, 7 December 2017 (UTC)
  •   Comment It is reasonable to politely dispute the edits of a contributor and to bring them up for discussion, firstly with the user, and possibly after with the broader community. It is unreasonable and provocative to call that editing vandalism, especially for an established editor. I also would put it to you, that it is counter-productive to your argument and case to call it vandalism. (The word vandalism has other connotations and none of us is likely to agree with that claim, and accordingly give the discussion a wide berth, thereby not resolving your issues.) So I would suggest that you look to promote a discussion that defines the problem, and the possible solutions, and to do that without a blame approach. [Play the ball, not the person.]  — billinghurst sDrewth 19:31, 7 December 2017 (UTC)

Textiles

Wikipedia articles generally discuss a fiber, the yarn made from it, and fabrics knitted or woven from this yarn one article. As a result, Wikidata has items such as wool (Q42329) which is both <subclass of> animal fiber (Q874142) and of string (Q11064354) which is ultimately <subclass of> artificial physical object (Q8205328). This is ontologically problematic. I propose that we should have wool1 = a fiber, wool2 = yarn or thread made from wool fiber, and wool3 = fabric woven from wool yarn. I am willing to make these changes (over time!) (including disambiguating <material used> statements). I'd like to be sure others agree with this approach before I begin (and if anyone is interested in forming a project on textiles to help, let me know). - PKM (talk) 20:26, 7 December 2017 (UTC)

  Support When we have wiktionary integration we'll probably want wikidata items associated with each definition (or at least some reasonable approximation of that - perhaps some definitions are too close to require a separate item) for words like this that may be ambiguous. wool has 6 definitions, so your 3 is a good start at least. ArthurPSmith (talk) 20:45, 7 December 2017 (UTC)

WiDaR and OAuth issues

For me trying to use toollabs:mix-n-match I am having OAuth authentication quirkiness — partial errors, says logged in, though not showiong user name. Anyone else having issues?  — billinghurst sDrewth 23:57, 12 December 2017 (UTC)

Seems that there was a server issue, resolved now.  — billinghurst sDrewth 03:38, 13 December 2017 (UTC)
This section was archived on a request by: Matěj Suchánek (talk) 17:50, 14 December 2017 (UTC)

What is the best way to fix One-to-many (Q7092345)?

I found an item One-to-many (Q7092345). It's marked as a Wikimedia disambiguation page (Q4167410), but both of the site links lead to articles about a relational data model. There is a separate item for a disambiguation page of the same title (One-to-many (Q16880738)). What is the best way to fix One-to-many (Q7092345)?

  1. Remove all the descriptions and repurpose the item as a data model? If so, is there a tool that could get rid of all the existing descriptions in that item?
  2. Don't repurpose the item, instead create a new item, move the site links there, and delete Q7092345?

Shinnin (talk) 19:43, 13 December 2017 (UTC)

I don't really see the purpose of not repurposing disambiguation items. Those items are "Wikimedia-internal", so to speak, and shouldn't have any external uses. If they do have external uses, it's most likely because someone grabbed the item number from the connected Wikipedia articles, in which case "not repurposing" will actually bring about the kind of damage the rule intends to avoid. For normal items I absolutely agree with the rule, but for Wikimedia-specific items I don't see the point in being rigorous with that rule. Jon Harald Søby (talk) 20:47, 13 December 2017 (UTC)

It seems that someone already created a new item and made the suggested redirect. The new item is one-to-many (Q45406573). Thanks for your input. I'll mark this topic as resolved. Shinnin (talk) 23:26, 14 December 2017 (UTC)

This section was archived on a request by: Shinnin (talk) 23:26, 14 December 2017 (UTC)

Men national team

Hello. It was discussed again (a little) but no solution... Is there a way with a query to find all men national football teams? No women teams, no U21, U19 etc teams. Just men teams. And if it is not, suggest a way to do this (by adding property for example). 15:48, 2 December 2017 (UTC)

One suggestion was to use competition class (P2094) --> men's association football (Q35554451). Is that acceptable from the community? Xaris333 (talk) 16:58, 2 December 2017 (UTC)

As far as I remember, I suggested this solution or at least strongly supported it. I still do so :-) —MisterSynergy (talk) 17:08, 2 December 2017 (UTC)
It is solving the problem with men/women teams but is not solving the problem with U23, U21, U19 etc teams. For those teams we can use under-23 sport (Q14510042), under-21 sport (Q21152744), under-20 sport (Q21198418) etc. But should we use an item "senior sports"?

And how to use competition class (P2094)?

or

Xaris333 (talk) 17:23, 2 December 2017 (UTC)

—MisterSynergy (talk) 17:49, 2 December 2017 (UTC)

@MisterSynergy: Ok. So, all U21 teams must have “men's U21 association football”, all U19 “men's U19 association football” and all senior teams nothing? Like this way:

and

and

and

Why we need two items for U19 teams? Why not to have one "U19 association football"?

Xaris333 (talk) 18:18, 2 December 2017 (UTC)

You don’t need both men's U19 association football (Q44408003) and men's association football (Q35554451) for U19 teams, the former is sufficient (same for women and other age classes). You can expect items to have only one P2094 claim, which carries information about both gender and age (as men's U19 association football (Q44408003) does). —MisterSynergy (talk) 19:37, 2 December 2017 (UTC)
Tried to make it clearer by striking unnecessary claims in your comment. —MisterSynergy (talk) 19:38, 2 December 2017 (UTC)

Thanks. It is clear now. Xaris333 (talk) 20:07, 2 December 2017 (UTC)

@MisterSynergy: There are some special cases of national teams that I want your opinion how to use competition class (P2094).

Xaris333 (talk) 18:17, 8 December 2017 (UTC)

Interesting to learn what kind of “national teams” exist. For the first two cases I would use the same P2094 value as with regular national teams in any way, and probably for the other two items as well. Disambiguation should rather be done with a different value for P31. —MisterSynergy (talk) 19:44, 8 December 2017 (UTC)

@MisterSynergy: Like what? "national B association football team", "national C association football team", "national amateur association football team", "national military association football team"? Xaris333 (talk) 20:04, 8 December 2017 (UTC)

The last two for amateurs and military personnel sound good, but I guess you only need one item “secondary national association football team” for the first two. It does not really matter whether it is call “B”, “C”, or whatever. They can all subclass national association football team (Q6979593), or, in case you disagree with that, association football team (Q15944511). —MisterSynergy (talk) 20:12, 8 December 2017 (UTC)
@MisterSynergy: Why can we use this new items with competition class (P2094)? All are national teams so instance of (P31)-->national association football team (Q6979593). So, if I want to search for all national teams of Spain, it will be easier to search for instance of (P31)-->national association football team (Q6979593) with country for sport (P1532)-->Spain (Q29). And P2094 would show me any possible national association football teams of Spain exist. Xaris333 (talk) 20:18, 8 December 2017 (UTC)
They should all have the same P2094 as all other national team items, but P31 should be different. If the subclass relation subclass of (P279) national association football team (Q6979593) is in place in the new items as suggested in my previous comment, this can very efficiently be queried:
This works regardless of all other claims in the items. Mind that items with subclass of (P279) national association football team (Q6979593) inherit the character of national association football team (Q6979593), thus they describe national teams as well.
Anyway, you’ve raised a good question. Whenever one sets something up as in this case, one of the important aspects to consider is: can it be queried efficiently? Answer here is: yes it can :-) —MisterSynergy (talk) 20:47, 8 December 2017 (UTC)

Ok, thanks. I will apply it. Xaris333 (talk) 20:50, 8 December 2017 (UTC) 20:49, 8 December 2017 (UTC)

how can you question a wikidata statement that seems incorrect?

On this wikidata item Andrea Zittel (Q495290) there is a statement that the person speaks or writes Italian. When you follow the reference, it appears only to show that a book is written about this person in Italian, not the person speaks Italian. How does one question a statement in wikidata? Thank you.  – The preceding unsigned comment was added by Michelleif (talk • contribs) at 19:18, 6 December 2017‎ (UTC).

If the reference doesn't support the statement, it can be removed. Then if someone questions the statement itself, it can be removed as well. --MB-one (talk) 21:12, 6 December 2017 (UTC)
Prod the person who added it; or undo its addition, with a comment "not supported by references". It can be readded with refs if necessary.  — billinghurst sDrewth 03:23, 7 December 2017 (UTC)
If there was a reference supporting it, it should be deprecated rather than deleted. Deprecation avoids that it gets re-added.
--- Jura 07:09, 7 December 2017 (UTC)

In the past I have considered adding sourcing circumstances (P1480) = citation needed (Q3544030), but someone removed this saying it wasn't in the controlled vocabulary for P1480. That was a shame, I think, because the "citation needed" mechanism has been very valuable on the Wikipedias for this kind of issue, as well as quite a mascot for the movement. IMO it would be good to have something similar here. Jheald (talk) 10:24, 7 December 2017 (UTC)

@Jheald: Having a citation needed mechanism would be nice, but that doesn't mean that you can simply put it into a property that was designed for a different purpose. ChristianKl (✉) 12:42, 7 December 2017 (UTC)
It seemed to me that actually "citation needed" fits in quite nicely with sourcing circumstances (P1480), which can currently be used to flag a rather random set of things 'not quite right' with an item, eg approximate location / approximate date / suspected misprint / suspected miscalculation / unclear calendar / etc.
As to reason for deprecated rank (P2241) (ping @Jura1:), that presumes that the statement should be deprecated. It seems to me that's not always the case. Sometimes the statement may be reasonably plausible, may even have some sourcing (eg to a wiki), and probably one doesn't want to see it left out of a 'truthy' search. But nevertheless, as a non-specialist, one still might want to record that one would want to see some more evidence and/or reasoning to support or explain it. Jheald (talk) 15:55, 7 December 2017 (UTC)
My experience of "citation needed" is that it usually means "I think this is probably true, but we need to prove it", rather than "I think this is wrong but I'm not sure" - and making a statement deprecated is very much "I think this is wrong". I think a property to allow tagging contested-but-not-necessarily-wrong statements is a very good way to go, whether it's P1480 or something else. Andrew Gray (talk) 13:01, 8 December 2017 (UTC)
  •   Comment It would seem that a help section in one of our primary pages that has a primer addresses "when to remove a statement versus when to deprecate a statement" would be useful guidance to our contributors, then with a link to Help:Deprecation for full information about that component.  — billinghurst sDrewth 22:33, 8 December 2017 (UTC)

Controller extension/accessory?

I would like to specify that Nunchuk (Q4894974) is an optional extension/accessory to Wii Remote (Q56898). How should I do this? I was hoping that I could do something like Nunchuk (Q4894974) instance of (P31) [hardware extension (or something similar)] of (P642) Wii Remote (Q56898). Pizza1016 (talk | contribs) 08:03, 1 December 2017 (UTC)

Anyone? Pizza1016 (talk | contribs) 01:48, 4 December 2017 (UTC)
It isn't generally useful to put qualifiers on P31 statements. You might want to claim Nunchuk (Q4894974) part of (P361) Wii Remote (Q56898) but that's not something that was always true, right? Generally in a case like this I would check out the original wikipedia article (maybe in more than one language) and see what statements there reference other articles which should have wikidata items. Maybe there's nothing that fits "hardware extension" (or "Wii Remote extension") and you will need to create a new wikidata item for those for this purpose. ArthurPSmith (talk) 19:20, 4 December 2017 (UTC)
Well, I created Wii Remote extension (Q44753009), but the subclass of (P279) component (Q1310239) statement may not be the best. - Pizza1016 (talk | contribs) 04:46, 10 December 2017 (UTC)

Possible confusion over name of Property Te Papa agent ID (P3544)

I thought I should let folk know of a possible issue with the title of the property Te Papa agent ID (P3544). The Te Papa Museum identifier relates not just to artists whose works are held at the museum but also to scientists, collectors and other people who have contributed to the Museum of New Zealand such as curators and donors. See for example the identifier used in this item Jagoba Malumbres-Olarte (Q21390767) relating to a scientist. As a result I believe the property should be renamed as "Te Papa ID" to avoid confusion. --Ambrosia10 (talk) 20:33, 8 December 2017 (UTC)

I moved this conversation to Property talk:P3544 and I'll reply there. Multichill (talk) 13:46, 9 December 2017 (UTC)

Monument listing property?

Hello, is there a field/claim/property for listed monuments? If so, does it allows for more information, such as the grade, source and year of listing? It would be extremely helpful to anyone organizing WLMs and similar events.--DarwIn (talk) 01:33, 9 December 2017 (UTC)

Hi DarwIn,
the most fundamental property in that domain is heritage designation (P1435). Then there are many properties for the monument IDs, such as World Heritage Site ID (P757), Berlin cultural heritage ID (P2424) or Canadian Register of Historic Places ID (P477). For more specific questions, Wikidata:WikiProject Cultural heritage is a great resource.
--MB-one (talk) 11:33, 9 December 2017 (UTC)
That project got moved to Wikidata:WikiProject Built heritage. On Wikidata:WikiProject Built heritage/Wikidata properties I put a (rough) query to get the different types and properties. Multichill (talk) 13:42, 9 December 2017 (UTC)
Hi @MB-one:, @Multichill:, thank you very much for the heads up! Will definitely have a look on that in the future. Seems quite interesting and with a great potential, indeed.--DarwIn (talk) 23:50, 9 December 2017 (UTC)

What is the inverse of discontinued date (P2669)?

See the title... Since we use this for a broad collection of subjects, it may be hard to connect this with a current property. Sjoerd de Bruin (talk) 19:34, 9 December 2017 (UTC)

Q numbers instead of labels?

I am seeing instances of Qnnnn appearing in statements rather than their EN labels. (Example at Q43303600 where three statements show Qnnnn even though two of them have EN labels.). Is this just me? - PKM (talk) 20:03, 9 December 2017 (UTC)

I saw a P number when adding a statement earlier today. Maybe a late effect of the two bugs mentioned above.
--- Jura 20:15, 9 December 2017 (UTC)
I did a purge and the labels appear now. - Pizza1016 (talk | contribs) 23:28, 9 December 2017 (UTC)
Thanks, I see the labels now. - PKM (talk) 00:18, 10 December 2017 (UTC)

United States of America (Q30) properties

  Notified participants of WikiProject United States. The United States of America (Q30) being the country with most Wikidata properties, I try to maintain the templates or categories that list them all. So I have just initiated a list of properties by U.S. state (Q35657) on Template talk:United States of America properties. Don't hesitate to add more properties there if you find relevant ones. Thierry Caro (talk) 01:00, 10 December 2017 (UTC)

Item definition

When there are several conflicting descriptions of an item, which one is definitive? (please ping with reply) Pbsouthwood (talk) 03:58, 10 December 2017 (UTC)

  • @Pbsouthwood: In my opinion, the one that matches the statements on the item.
    --- Jura 08:21, 10 December 2017 (UTC)
  • @Pbsouthwood: I agree with Jura, when you have enough statements to define the item, that's the definition that's best. In practice there are unfortunately currently many items that don't have enough statements to define them well. In those cases it makes sense to add statements to the item that define it's meaning and create an additional item for any additional item that can be well defined with statements. If you feel like there's important information that you don't know how to express with statements, feel invited to come to the project chat to get ideas from other people of how to specify the information. If we currently don't have the necessary properties to express the information we can create new ones. ChristianKl (✉) 14:16, 10 December 2017 (UTC)
    Does the original description not take priority as the definition?
    If for example in one language the label is diving and the description describes diving and another the label means diver and in others both are diving and diver (in another language are used as aliases but there is no description, how does one know which was originally intended? The problem is that there may be a whole bunch of properties associated which do not obviously distinguish the intended meaning to a lay person, and if the wrong definition is selected,some of the properties will be wrong. In this case, on possibility is an activity and the other is the agent of the activity.
    I have changed a few labels in other languages where they disagreed with the English label, when the English label agreed with the English description, but I am concerned as this could be construed as overruling the original creator who may have provided a perfectly good label and description in a language I don't understand. I know would be annoyed if someone did that to an item I had created. Surely there is a system to deal with these cases?
    If I understand correctly, the item is ultimately defined by its properties, but this only works if there are no contradictory properties. It requires a knowledgeable person or an expert to distinguish which properties are applicable in some cases. Pbsouthwood (talk) 15:36, 10 December 2017 (UTC)
    The main statements would be P31/P279. Initial creation of items frequently imported any articles that were linked together by interwikis and (for some languages) many aliases which can contradict Wikidata logic (sample deletion on pascal (Q44395)). Once in a while items need to be split into new items as the original one just combined too many things (sample work-in-progress: Q2965940).
    --- Jura 15:47, 10 December 2017 (UTC)
    We currently don't have any policy that declares the original description as the description that's the most important one. I think there's a valid argument to be made for doing so, but at the moment the best step forward is to split the item into two. Hopefully as time goes on different people we clarify the meaning of examples like diving and diver and clean up those items. ChristianKl (✉) 19:14, 10 December 2017 (UTC)
    OK, I guess I will just have to play it by ear. Thanks anyway,Pbsouthwood (talk) 05:59, 11 December 2017 (UTC)

possible problem with recent edits on Polje (Q7225446)

Could someone who knows geography of Bosnia-Herzegovina please investigate the recent edits (and link removals) on Polje (Q7225446).

The removals seem suspicious to me, as well as the link from srwiki... but I don't know enough about the subject to be able to tell really... --Hsarrazin (talk) 13:04, 10 December 2017 (UTC)

I still don't know why people do this. Sjoerd de Bruin (talk) 13:12, 10 December 2017 (UTC)

P31 and P279 containing exactly the same value

myostatin (Q29630).


Only one can be true. 77.180.202.181 14:08, 10 December 2017 (UTC)

I don't see an issue arising from the way the properties are used there. ChristianKl (✉) 19:16, 10 December 2017 (UTC)

Wikidata Liaison for WikiJournals

Hello all,

I'm involved in the WikiJournal user group (Wikipedia-integrated academic journals; see refs below for more details). WikiJMed is the most established if you want to look at examples. I feel that there are likely great things that can be done in collaboration with Wikidata, but I'm sufficiently novice to not know the possibilities. Some half-formed ideas from a thread on FB's Wikipedia Weekly:

  1. add metadata for WikiJournal articles automatically to WikiData
  2. integrate with data presentation tools like Scholia
  3. broaden with some ResearchGate-like manuscript archiving system using wikimedia commons
  4. add AltMetrics or similar badges (currently available only for whole journal, not individual articles)

I'm therefore looking for a Wikidata Liaison to advise / mediate / work with the current WikiJournals to help with ideas and implementation. What do people reckon?

More info:

Evolution and evolvability (talk) 12:18, 8 December 2017 (UTC)

@Evolution and evolvability: I'm happy to help, but I don't think you need a single=-person liaison. If you do (1) in your list above then (2) will follow automatically. Also, we have Wikidata:WikiProject Periodicals, Wikidata:WikiProject Source MetaData and m:WikiCite which may be useful to you. Andy Mabbett (Pigsonthewing); Talk to Andy; Andy's edits 12:38, 8 December 2017 (UTC)
@Pigsonthewing: You're right of course that a single person isn't really a necessity and that the same can be achieved in a distributed manner across multiple people. I'll have a look into the links you suggested and get back once I've got my thoughts in order (still so much to learn about wikidata!) Evolution and evolvability (talk) 00:20, 11 December 2017 (UTC)
One initial example idea, would it be possible to make a tool that allows people to generate a unified Wikimedia login based on their ORCID ID (analogous to how eLife allows people to log in using their ORCID account). It could even have an option to auto-generate a userpage with basic data from their ORCID page? Evolution and evolvability (talk) 02:44, 11 December 2017 (UTC)
Technically possible yes, but at a cost. See meta:ORCID for some thoughts on possible future integration. Andy Mabbett (Pigsonthewing); Talk to Andy; Andy's edits 10:56, 11 December 2017 (UTC)

Contributions do not update

Hoi, it has been since November 27 that my last update happened. Do we have a problem ? Thanks, GerardM (talk) 15:11, 11 December 2017 (UTC)

Can you be more precise? Sjoerd de Bruin (talk) 16:07, 11 December 2017 (UTC)

merging entries

Sorry, I am really new to Wikidata...

I wanted to update this entry : La Nouvelle Liberté (Q21094001)

But then found it was not possible because of this one : La Nouvelle Liberté (Q3211132)

The system suggests merging entries. Good move. But... is it something we simply do by hand ? Or is there a way to merge them with a tool ? How would that work ? Please do not do it for me. Explain me if I can do it myself and if so how.

Thanks

Anthere (talk)

In this case merge isn't the correct gadget, better use move and propose it for deletion, they are different type of item. --ValterVB (talk) 16:28, 7 December 2017 (UTC)
ok thanks Anthere (talk) 19:58, 7 December 2017 (UTC)
@Anthere: there are also two tools to move bits, one for interlinks and another for statements on an item.  — billinghurst sDrewth 22:23, 8 December 2017 (UTC)
@billinghurst: hi. Where are those tools ? Anthere (talk) (nota bene : someone took care of merging the two)
@Anthere: One is a gadget, and the other is User:Matěj Suchánek/moveClaim.js.  — billinghurst sDrewth 20:12, 11 December 2017 (UTC)

"mayor of X" or "Mayor of X"

"mayor of X" or "Mayor of X" for the position? I am looking at all presidents, governors, and mayors and there is a mix of both. --RAN (talk) 20:28, 10 December 2017 (UTC)

In my view "mayor of X" should be the position (i.e. it should be an instance of position, or an instance of subclass of position, etc.) It should not be the person, and should not be a subclass of eg "politician". It would be good to clean this up systematically across the site. Jheald (talk) 15:16, 11 December 2017 (UTC)
The capitalized version IMHO is the formal title and the uncapitalized as the generic position. The Mayor of New York City, John Lind, has been serving as the mayor of New York City for 6 years. Standardization would be nice. --RAN (talk) 01:43, 12 December 2017 (UTC)

Wikidata weekly summary #290

 
Here's your quick overview of what has been happening around Wikidata over the last week.
  • Other Noteworthy Stuff

You can see all open tickets related to Wikidata here.


Editing labels and descriptions

Sometimes when I click the edit button to change the labels and descriptions of an item, a box appears with editable fields and sometimes I am redirected to the Special:SetLabelDescriptionAliases page and I can edit only one language. I don't understand the rule for this, and I don't know how can I reach the first interface in the second case (I would like to edit an other language, but there is no option for that on this surface). Samat (talk) 21:47, 11 December 2017 (UTC)

I think you end up on the "Special page" if you click "edit" before the page has been completely loaded. To get back just click on the back button on your web browser and try again. --Larske (talk) 22:02, 11 December 2017 (UTC)
Thank you for that explanation! - PKM (talk) 02:50, 12 December 2017 (UTC)
Thank you! It looks for me, that you are right. Samat (talk) 08:15, 12 December 2017 (UTC)

Changing the property

Sometimes there is a big statement group with statements, qualifiers and references, and I should change the property (wrong or deprecated).

  • Is there a way to change the property instead of copying by hand (filling up the separate fields one by one) the whole statement group?
  • Is it possible to copy a whole statement group to a new place (e.g. into a new item, based on a good example)?
  • Does Wikidata have a "wikicode" mode / editing surface maybe? :)

Samat (talk) 21:55, 11 December 2017 (UTC)

Some bots can do that. @Pasleim:. Xaris333 (talk) 22:16, 11 December 2017 (UTC)

  • "move whole statement group to different property" is only possible with bot (or pywikibot code in PAWS), as Xaris333 correctly stated
  • whole statement groups can be copied (or moved) to different items using moveClaim.js by User:Matěj Suchánek (put mw.loader.load( '//www.wikidata.org/w/index.php?title=User:Matěj_Suchánek/moveClaim.js&action=raw&ctype=text/javascript', 'text/javascript' ); // [[User:Matěj Suchánek/moveClaim.js]] to Special:MyPage/common.js, then you’ll see an extra button in each statement)
  • there is no "wikicode" or "wikisource" more for source-based manual editing available

—MisterSynergy (talk) 14:39, 12 December 2017 (UTC)

Thank you! Samat (talk) 17:47, 12 December 2017 (UTC)

Where should problems with infoboxes that use Wikidata content by directed?

I just get access to the OTRS queue and one of the requests is from a person who had trouble with an EnWiki infobox that uses Wikidata content and throws Lua errors. Do we have a place towards which we can direct a person with such a request? ChristianKl (✉) 13:30, 12 December 2017 (UTC)

The person must write the problem to a page to enwiki (template's talk page or w:en:Wikipedia:Village pump (technical)). It is not something that can be solved in wikidata. The problem, I guess, has to do with the modules of enwiki. Xaris333 (talk) 14:00, 12 December 2017 (UTC)

Freedom in the World (Q1417780) status for countries?

This status (free, partly free, not free) is often used by NGOs or activist groups. For a project I'm working on it would be very useful to be able to filter countries according to it, but I can't figure out the best way to add this information. Things like review score (P444) and ranking (P1352) aren't quite right. Is there an existing property that would be more suitable? --Oravrattas (talk) 08:09, 12 December 2017 (UTC)

Maybe "has quality" "free according to Freedom House"? ChristianKl (✉) 12:02, 12 December 2017 (UTC)
Oh, nice idea. has characteristic (P1552) could definitely work here. Though perhaps using free country (Q3174312) with a statement supported by (P3680) qualifier of Freedom in the World (Q1417780) (or Freedom in the World 2017 (Q30539476)?) might be better than creating a specific "free according to Freedom House" item? --Oravrattas (talk) 16:20, 12 December 2017 (UTC)
I think in this case Freedom House not only supports the claim of it being free but also defines it. ChristianKl (✉) 18:47, 12 December 2017 (UTC)

Mineralogy

Hello. Just a small check. Is my view ok or controversial? Any ideas/comments?
Hierarchy: science > exact sciences > earth sciences > geology > rocks > mineralogy (minerals & nonminerals; mineral classification) > mineral classes & subclasses (Strunz) > mineral supergroups > mineral groups > mineral subgroups (mineral series), minerals, mineral polytypes & varieties.
Orientation (Classification)
Thank you very much --Chris.urs-o (talk) 18:04, 12 December 2017 (UTC)
Are you trying to build a subclass tree? Because rocks, the structures, are quite certainly not a subclass of geology, the scientific discipline. --Yair rand (talk) 19:10, 12 December 2017 (UTC)
I was just thinking. Well, geology studies the Earth crust, I have to think about that. Earth's crust > geological unit > geological formation > rock. But I am working on the tree below mineralogy. Thx --Chris.urs-o (talk) 19:38, 12 December 2017 (UTC)
@Chris.urs-o: As Yair rand suggests, your > means different things here - sometimes it would be a subclass of (P279) relation (from right to left), sometimes it might be facet of (P1269), for some of these things instance of (P31) might also be appropriate. Also what you propose is not entirely consistent with what we have now in wikidata - for example at the moment mineral subclass (Q3965272) subclass of (P279) mineral group (Q1936581) while it appears from your list this relation would be in the other direction. I'm not familiar enough with this field to say what's right here. I suspect that most (all?) of your proposed specific mineral classes also already exist in Wikidata - for example sulfide class of minerals (Q927407) - so I would caution you to first look at what already exists here in these areas. If you want to make changes to what's been done already, you can come here to seek consensus, but of course also include Wikidata:WikiProject Mineralogy of which you seem to be a participant already. ArthurPSmith (talk) 19:46, 12 December 2017 (UTC)
Okidoki. Mineral subclass subclass of mineral group is wrong. No panik, I am not changing anything major. Thx
Note: I (124k), Tobias1984 and Sbisolo (in the beginning) did many edits on Wikidata:WikiProject Mineralogy. --Chris.urs-o (talk) 21:08, 12 December 2017 (UTC)

Recoin2: New version of relative completeness status indicator for all Wikidata entities

Dear community,

We are happy to announce version 2 of Recoin, the relative completeness status indicator. Recoin adds a color-coded progress bar to entity pages showing how extensive information about the entity is in comparison with similar entities (see example picture for the album Abbey Road).

 
Recoin on Abbey Road

In version 2, Recoin is rolled out to virtually all entities in Wikidata (see e.g. Q565400 or Q15074414 ). The core version uses only Wikibase properties, but based on earlier feedback there is also a version specifically for ID properties only.

We think Recoin can be of great help for quickly getting impressions of data quality about entities. We look very much forward to your feedback.

Also, if you support making a gadget out of it please get in contact with us.

More information can be found at Wikidata:Recoin.

Best regards,

Ls1g (talk) 08:20, 6 December 2017 (UTC)

Good news Ls1g! I'm happy to see you moved to the Wikimedia Toolforge, that takes care of the privacy concerns. The project seems to be maturing. Time to turn it into a gadget?
Maybe also time to move it to Wikidata:Recoin? Multichill (talk) 17:32, 6 December 2017 (UTC)
Thanks for the hints! (2) Just moved it to Wikidata:Recoin. (1) Gadget: That would be great - can you advice us what would be the process? Ls1g (talk) 11:08, 7 December 2017 (UTC)
Sadly had to deactivate it again, yesterday it worked fine, but today it slowed down the webpage so much it became barely usable. Does it calculate the completeness after every edit, or only at initial page load, or does some caching? I noticed that properties I added already were still listed among those missing. Ahoerstemeier (talk) 09:49, 7 December 2017 (UTC)
Can you describe the issue further? For me it does not affect the webpage loading, the page loads fast as normal, then a second or two later the indicator is added. Regarding computation, it is done only once, at page load, and uses the SPARQL endpoint. So to see changes the page has to be reloaded (and the SPARQL endpoint has new changes usually after 1-2 minutes). Ls1g (talk) 10:30, 7 December 2017 (UTC)
Update: My colleague just told me it seems something can be optimized regarding the speed of the underlying DB queries, I'll report later. Ls1g (talk) 11:08, 7 December 2017 (UTC)
It was not the page loading which was slow, it was the time needed when adding a statement until I could select the property or type anything. As it went back to normal directly after deactivating Recoin I guess that was the culprit - though a few sections below others report a similar slowness which now recorded as Bug T182322. Ahoerstemeier (talk) 14:27, 7 December 2017 (UTC)
@Ahoerstemeier, Ls1g: There is comment below about page loading slowness as a universal issue. It may be coincident of issues.  — billinghurst sDrewth 22:35, 8 December 2017 (UTC)
Thanks for pointing this out! We also did some optimizations on our side, now Recoin should load considerably faster. Ls1g (talk) 13:07, 11 December 2017 (UTC)
Thanks, I already activated it again and it works like a charm now. And nice to see items I care for to have a green indicator. Ahoerstemeier (talk) 09:09, 13 December 2017 (UTC)

@Ls1g: Hi, I work mainly on "people" items (Q5), and for newly created items, which only have P31... it displays a list if IDs (VIAF, LCCN, etc. -- see screenshot, but NOT sex or gender (P21), country of citizenship (P27), date of birth (P569), date of death (P570), image (P18), occupation (P106), + name in native language (P1559), family name (P734)/given name (P735) that are the most essential properties for people... why is that ? --Hsarrazin (talk) 09:45, 13 December 2017 (UTC)

@Hsarrazin: Hi, please use the other script (recoin.js) - there are two variants of Recoin, recoin.js shows "normal" properties, which is what you and presumably most people want, whereas recoin_id.js, the one you have currently enabled, is a derivative that only shows IDs like VIAF etc. Thanks for pointing this out, I'll fix the description to make the distinction more clear. Ls1g (talk) 13:00, 13 December 2017 (UTC)
@Ls1g: ahhhh... thanks for this piece of info :D --Hsarrazin (talk) 13:15, 13 December 2017 (UTC)

Original research in the Wikidata?

In Wikipedia, you can make a statement and indicate its source. For example, w: Musician: "A musician (or instrumentalist) is a person who plays a musical instrument or is musically talented. [1]". In Wikidata, this possibility is absent for some cases, which allows you to create original research here: musician (Q639669) (activity of a person) is both profession (Q28640) (person) and performing artist (Q16010345). Any references to such statements can not be given now. --Fractaler (talk) 13:19, 7 December 2017 (UTC)

Fractaler, fixed by removing the false claims from "musician" [9] 92.226.213.195 14:11, 7 December 2017 (UTC)

I undid the vandalism from the IP user. Please do not mess around with wikidata ontology like that. As to Fractaler's question, there's no reason why references cannot be given for instance and subclass statements; generally they are considered to be obvious and not require references, but if references are needed here I think for example this one which states (1) a person who makes music a profession, especially as a performer of music as a definition works. ArthurPSmith (talk) 15:20, 7 December 2017 (UTC)
The one who is messing is ArthurPSmith. Edit by idiot rreverted, not every musician is a performer. Read enWP "Anyone who composes, conducts, or performs music may also be referred to as a musician" 213.39.186.16 15:29, 7 December 2017 (UTC)
So, here we have: musician (Q639669) is a person. --Fractaler (talk) 18:53, 7 December 2017 (UTC)
Actually, the wikidata item in question refers to a *class* of people. An instance of musician (Q639669) would be a person. musician (Q639669) itself is one of many conceptual classifications of people (like "scientist", "entrepreneur", etc) that are generally described as professions, occupations, hobbies, skills, etc. ArthurPSmith (talk) 19:42, 7 December 2017 (UTC)
As profession? Link, please. Or: original research.--Fractaler (talk) 19:49, 7 December 2017 (UTC)
Link above. ArthurPSmith (talk) 20:21, 7 December 2017 (UTC)
Link above said: "musician: a person who ...". --Fractaler (talk) 06:34, 8 December 2017 (UTC)
No different from lawyer (Q40348) which is also an instance of profession (via a subclass), and defined in same source as a person whose profession is to represent clients in a court of law or to advise or act for clients in other legal matters. All professions, occupations, etc. in wikidata are like this: classes whose instances would be individual persons, but the class itself is an instance of "profession" (or one of the other similar classes). ArthurPSmith (talk) 14:36, 8 December 2017 (UTC)
Yes! Source said: person, but Wikidata said: profession. What this original research are doing in the Wikidata?--Fractaler (talk) 18:06, 8 December 2017 (UTC)
Source also said "profession". ArthurPSmith (talk) 18:15, 8 December 2017 (UTC)
Note the enwiki quote above says "A musician is a person..." i.e. an instance of the class musician is also an instance of the class person. But it doesn't say anything about what the class musician itself is (other than a way of filtering persons). That's what professions, occupations, etc. are - classes of people. ArthurPSmith (talk) 18:24, 8 December 2017 (UTC)
Source don't said: a musician is a profession. Ok, if professions are classes of people, then what is their activity, their professional occupation?--Fractaler (talk) 18:51, 8 December 2017 (UTC)
I can't find many wikidata items for the activity of a professional/occupation, but here's one: teaching (Q28545). For a musician I suppose it would be "making music". art (Q735) more generally encompasses what musicians do. ArthurPSmith (talk) 19:28, 8 December 2017 (UTC)
Wikidata's original research said: 1) musician (Q639669) -> profession (Q28640) -> occupation (Q12737077) -> activity (Q1914636) 2) lawyer (Q40348) -> legal profession (Q15987302) -> profession (Q28640) -> occupation (Q12737077) -> activity (Q1914636), etc. musician (Q639669) is activity (Q1914636)? lawyer (Q40348) is activity (Q1914636)? Or musician (Q639669) (lawyer (Q40348), etc.) is human (Q5)? --Fractaler (talk) 14:07, 9 December 2017 (UTC)
Ah, so perhaps occupation shouldn't be a subclass of "activity"? I think you have a case there. ArthurPSmith (talk) 15:55, 9 December 2017 (UTC)
But now WD said: occupation (Q12737077): "any activity of a person". --Fractaler (talk) 16:13, 9 December 2017 (UTC)

┌────────────────────────────────────────────────────────────────────────────────────────────────────┘ I agree having occupation as a subclass of activity is a problem. “Singing” is an activity. “Singer” is not. The question then becomes what is an occupation? Common sense says a singer is a person who sings. Getty AAT says painters are “People who practice the art of painting”. Since we seem to have agreed that human (Q5) shouldn’t have subclasses (although it has a few), that simple logic doesn’t work here. Is occupation some sort of role? Is it a “description of a person based on activities that person engages in”? An “aspect of human”? - PKM (talk) 19:52, 9 December 2017 (UTC)

@PKM, ArthurPSmith, and Fractaler: I found "Wikidata:What Wikidata is not" guideline saying that original research is disallowed... or that Wikidata is "not a place to put original research". George Ho (talk) 04:03, 11 December 2017 (UTC)
This isn’t really about original research. No one is disputing the “fact” that a musician is a person. The question is how that fact should be represented in our particular ontology. - PKM (talk) 04:40, 11 December 2017 (UTC)
WD now has: "human+human's activity" (what is it?). Moreover, what kind of activity? catalytic activity (Q14353133) (catalytic activity), gastric activity, volcanic activity, etc. also is a activity? As far as I understand, organism (Q7239) is the carrier of either innate (inherited characteristics) or acquired properties (like acquired characteristic (Q4674771)). The professional/occupation/etc. is similar to the acquired property, but certainly not innate. What is a profession, occupation? It is some skill (Q205961) (acquired characteristic) +/- document on this +/- place of activity? So, human (Q5) can be native speaker (Q924130) (language's carrier), profession/occupation's carrier, etc. And if musician (Q639669) is human (Q5) as "carrier of acquired properties" (and only), then we need "musician (Q639669) (occupation)". And then musician (Q639669) has acquired characteristic (occupation (P106)) "musician (Q639669) (occupation)". Fractaler (talk) 07:53, 11 December 2017 (UTC)
profession (Q28640) has job function (Q828721)? human (Q5) has job function (Q828721)? --Fractaler (talk) 14:09, 13 December 2017 (UTC)

Retracted paper

Hoi, I created a qualifier for "retracted paper". I added to one of the more infamous papers. Please comment and we can start add this to for instance what "Retraction watch" has to say and we can include it where a template for retracted papers is used on en.wp. Thanks, GerardM (talk) 13:29, 10 December 2017 (UTC)

Try constructing something with significant event (P793) instead of this constraint violating construction. Sjoerd de Bruin (talk) 13:48, 10 December 2017 (UTC)
That works for me.. Should this not be higher up in the ranking for the item? Thanks, GerardM (talk) 13:57, 10 December 2017 (UTC)
If you use significant event than the item should be named something like "retraction of the paper" and subclass event. ChristianKl (✉) 14:19, 10 December 2017 (UTC)
Wikidata is setup for different data users being able to display data in any order they like. The order in which properties are listed shouldn't be construed as reflecting a rank. ChristianKl (✉) 19:44, 10 December 2017 (UTC)
Why should it be a subclass of a scientific paper. It still is; it is only retracted. I do not understand your notion about ranks here. Thanks, GerardM (talk) 20:45, 10 December 2017 (UTC)
I thought that you meant with "rank" that you have a problem that it's lower down the page. If that's not the case, then that's great. If your argument is that about it being a "scientific paper" I don't think something stops being a scientific paper when it's redacted. We have external ID property that have type constraint that they can only be used with "scientific paper". Those don't suddenly become invalid when the paper gets redacted. ChristianKl (✉) 00:15, 12 December 2017 (UTC)

The Source MetaData WikiProject does not exist. Please correct the name. --Succu (talk) 23:01, 10 December 2017 (UTC)

I was pointed out to a recent paper that would help us by providing a differentiating in three classes. Have a read. Thanks, GerardM (talk) 15:09, 11 December 2017 (UTC)
The article seems to argue that it would be good if there are amendments and corrections to published articles. Wikidata won't host either and our role is to describe what the journal in question did with their article. If they say they retracted it, we should be calling the process retraction. If they say they amended it, we should call it amendments. Furthermore it would be interesting to have a look at the metadata that the journals use to say that they retract or amend a paper. ChristianKl (✉) 12:52, 13 December 2017 (UTC)

Professor

Does anyone have any ideas as to how to better disambiguate professor (Q121594) and full professor (Q25339110)? I suspect professor (Q121594) should be the profession and full professor (Q25339110) should be the academic rank, and therefore not a subclass of university teacher (Q1622272) (i.e. not a profession). I think the ENwiki links perhaps need to be swapped as well. Any academics want to weigh in? - PKM (talk) 20:31, 10 December 2017 (UTC)

This is completely mixed. For example, the Russian article combines both meanings.--Ymblanter (talk) 20:33, 10 December 2017 (UTC)
Please explain how professor is a profession imho it is only a rank. Thanks, GerardM (talk) 07:01, 11 December 2017 (UTC)
In some countries exists position, that sounds "professor". Maybe that led to misunderstanding? - Kareyac (talk) 12:59, 11 December 2017 (UTC)
The United States Bureau of Labor Statistics has Standard Occupational Classification System (Q7598269) codes for a number of professors by area of practice: 25-1011 Professors, Accounting; 25-1022 Professors, Actuarial Science; 25-1072 Professors, Advanced Nursing; 25-1032 Professors, Aeronautical Engineering, and so on [10]. I think that makes a pretty solid case for "Professor" being an occupation/profession in the US. - PKM (talk) 21:07, 11 December 2017 (UTC)

I think my problem with university teacher (Q1622272) vs. lecturer (Q1569495) related to this question. In my understanding first one is a person (and profession) who teaches on a university, second one is (the highest educational) position in a university. Every lecturer (Q1569495) is a university teacher (Q1622272), but only very few of university teacher (Q1622272) are lecturer (Q1569495). Samat (talk) 21:37, 11 December 2017 (UTC)

Looking at the English Wikipedia articles and English descriptions, I don't see why there are separate entries for professor (Q121594) and full professor (Q25339110). They both seem to be about the rank. We already have a separate entry and enwiki article for members of the profession: faculty member (Q5428874). Looking again at Samat's comment, I don't see why we have separate entries for faculty member (Q5428874) and university teacher (Q1622272). Again, they seem to be the same topic in English, though there may be a distinction in other languages. And we shouldn't have faculty member (Q5428874) has part(s) (P527) assistant professor (Q5669847). That's now how has part(s) (P527) should work. If anything, there should be a subclass of (P279) relation in the other direction. MartinPoulter (talk) 17:05, 13 December 2017 (UTC)
Professor means different things in different parts of the world. In the US and Canada a professor is anyone in a regular teaching position at a college or university. There are various ranks such as assistant professor and full professor. In some other countries professor refers only to those at the highest rank, the full professor. The English Wikipedia article en:Professor links to full professor (Q25339110) professor (Q121594) (whose description does not match the article, I will fix it) and en:Professor (highest academic rank) links to full professor (Q25339110) (whose description does match the article). StarryGrandma (talk) 20:15, 13 December 2017 (UTC)
Fixed links in the above. StarryGrandma (talk) 00:12, 14 December 2017 (UTC)

Feeding Wikidata into Wikipedia Infobox

This is in reference to an earlier query on contributing to Wikidata https://www.wikidata.org/wiki/Wikidata:Project_chat/Archive/2017/03#Contribute_entity_URI_to_Wikidata_2

We notice there are data which exists in Wikidata but are not exposed in Wikipedia infobox. How can we feed the data into Wikipedia infobox?

E.g. National Library Board (Q6974124) – The data “member of Biodiversity Heritage Library Consortium” exists in Wikidata, but not found in the infobox of Wikipedia article National Library Board (https://en.wikipedia.org/wiki/National_Library_Board). Is there a way to expose the data in Wikipedia infobox?  – The preceding unsigned comment was added by Nlbkos (talk • contribs) at 13. 12. 2017, 06:04‎ (UTC).

The infobox must support this data stream. You need to ask on the English Wikipedia whether they want to enable it. Matěj Suchánek (talk) 08:23, 13 December 2017 (UTC)
To be specific, if you want that something gets included into the authority control template of the English Wikipedia https://en.wikipedia.org/wiki/Template_talk:Authority_control is the place to raise the request. ChristianKl (✉) 19:09, 13 December 2017 (UTC)

Improving the data import process

 

Background:

The Request for Comment page and related discussion page linked to below follow on from the session held at WikidataCon 2017, which aimed to gather feedback from the community about problems/difficulties with the current data import flow. The main objective is to use this area to build a picture of what already exists and what's missing in the data import process. This can then be used to create actionable tasks (which we're proposing be managed on Phabricator).

What we want:

We want to get as many community members as possible involved in these early stages of planning. The end goal is to have a well structured task list showing the improvements we need to make to the data import process. We have created a draft page showing the general steps for the data import process, indicating the resources we have and what is still needed. We've also proposed a possible structure for a project on Phabricator, which also needs to be reviewed by the community.

If you have anything to add please go ahead and edit the page and/or join the discussion.

Wikidata:Requests for comment/Mapping and improving the data import process

Any help greatly appreciated! Thanks, NavinoEvans (talk) 10:22, 13 December 2017 (UTC)


The inclusion of the pages makes the TOC here rather big. Sjoerd de Bruin (talk) 10:51, 13 December 2017 (UTC)
@NavinoEvans: I would recommend that you create a Request for Comments for this discussion as that's a better way to deal with long documents than having them in the project chat. ChristianKl (✉) 11:51, 13 December 2017 (UTC)
I'll sort this out to be an RFC today. Thanks, --John Cummings (talk) 13:23, 13 December 2017 (UTC)
Changed to a linked RFC page, thanks. --John Cummings (talk) 20:08, 13 December 2017 (UTC)

Google often serves mobile pages for Wikidata even though I'm on a desktop

When I Google for Wikidata pages I often get m. mobile pages even though I'm on a desktop PC. Do other people find that the issue also exist for them? ChristianKl (✉) 15:32, 13 December 2017 (UTC)

oh yess, very often !! :(( - also on wikipedia and wikisource, but it's even more problematic on wikidata, due to the format being data and not text... --Hsarrazin (talk) 15:49, 13 December 2017 (UTC)
Do you use a mobile device while logged in to Google? AFAIK it remembers many things about your behavior and this could be a side effect. Matěj Suchánek (talk) 17:57, 13 December 2017 (UTC)
I never had this effect (I am sometimes using IPad but then choose the desktop version).--Ymblanter (talk) 18:12, 13 December 2017 (UTC)
I do use Google sometimes on my smartphone with uses the same Google account as my desktop computer but it would seem strange if Google isn't smart enough to separate the two uses. ChristianKl (✉) 19:07, 13 December 2017 (UTC)
There are several tasks for this in Phabricator, like phab:T60425 and phab:T72045. I find the fact that the mobile site doesn't redirect to the desktop site when you're on desktop extremely annoying, but the Foundation folks don't seem to agree. Maybe a few more voices on those tasks could have an impact? Jon Harald Søby (talk) 20:40, 13 December 2017 (UTC)

Help:Sources does not mention Wikisource

How do I reference a Wikisource page from an item? Also, shouldn't this basic need be explained in Help:Sources ? Jane023 (talk) 16:45, 9 December 2017 (UTC)

Jane, what do you mean? In an item, you can add a sitelink to Wikisource. Do you mean smth else?--Ymblanter (talk) 17:21, 9 December 2017 (UTC)
For example, this painting, A Rocky Landscape with Great Oaks (Q21015022), should reference de:s:Verzeichniss der von Speck’schen Gemälde-Sammlung, Gemälde Nr. 1 bis 89, since that is where the image comes from. Jane023 (talk) 17:45, 9 December 2017 (UTC)
Hmm and now I see that Woman in the window (Q30302574) is also in the same catalog. We probably have more. Jane023 (talk) 17:50, 9 December 2017 (UTC)
I know very little about Wikisource but I guess you can add it both as sitelink or as a reference.--Ymblanter (talk) 18:11, 9 December 2017 (UTC)
I did this for now. Jane023 (talk) 18:29, 9 December 2017 (UTC)
I thought that any Wikimedia project is not a good source by itself... May be better to use imported from Wikimedia project (P143)? --Infovarius (talk) 19:03, 13 December 2017 (UTC)
ah, but wikisource is not the source of the reference : the source is the edition of the work that can be found on wikisource :) --Hsarrazin (talk) 08:19, 14 December 2017 (UTC)
In references, I use “stated in” with the title of the edition and then a reference URL to Wikisource. - PKM (talk) 20:05, 9 December 2017 (UTC)
Ah that would mean making an item. Apparently these don't get automatically created for Wikisource editions? I would expect the landing page to have an item (or the index page). There is no item in this specific case. Jane023 (talk) 20:53, 9 December 2017 (UTC)
<sigh> I spend at least half of my referencing time making works and editions. - PKM (talk) 00:34, 10 December 2017 (UTC)
I am not against making an item but I just can't wrap my head around the distinction between works and editions. Maybe for Hamlet, but art catalogs? (throws up hands) Jane023 (talk) 13:09, 10 December 2017 (UTC)
Sounds good. We need to make sure these don't get replaced with Wikidata:Property proposal/Wikimedia reference link.
--- Jura 20:18, 9 December 2017 (UTC)

Football kit

  Notified participants of WikiProject Association football

Hello. I am using a lot the data of Wikidata to Wikipedia templates, especially about football. In templates about football teams we can have all the data from Wikidata except the football kit. I had proposed a property about that Wikidata:Property proposal/Home kit (football). I have understood why the proposal what not successful. So, I am writing here to listen to your ideas how we can have the football kit, as a data, to Wikidata. The idea is to be able to fetch the data from Wikidata so we can have the uniform it Wikipedia template. Please read the proposal first and then see how we are using the kid parameters in w:en:Real Madrid C.F.. I know that is not the real uniform, it is a representation of the uniform. But, its a very common way to show the kit of a football team in Wikipedia and we have all the pictures in commons. It would be easier if all Wikipedia could fetch the kit from Wikidata. Xaris333 (talk) 21:31, 11 December 2017 (UTC)

When creating a Wikidata property it's useful to think about how it generalizes. What other sports have things like a football kit? What's exactly the difference to an uniform? Is the kind of uniform that a Marching band carries something similar to a "football kit"?
Do you have good sources that describe how a football gets defined elsewhere? ChristianKl (✉) 18:59, 12 December 2017 (UTC)
@ChristianKl: Of course we can have a general representation uniform property. It can apply for many sports (I haven't see anywhere in Wikipedia representation uniform about anything else except sports). Well, this article explain (with sources) the football kit. Is that what you are asking? Xaris333 (talk) 20:44, 12 December 2017 (UTC)
When creating a property, it's good when it can not only represent one use-case but it can also represent multiple other use-cases. Currently, Navy Working Uniform (Q19297117) is for example linked to no other item but it would likely make sense to link it to another item about the Navy. If the relationship between this uniform and some other Navy item is qualitatively the same as the one between the Football kit and the football club, both use-cases should be handeled by the same property. To do that effectively it would be good to have a list of use-cases that the property should be able to handle well. ChristianKl (✉) 20:56, 12 December 2017 (UTC)
@ChristianKl: What about Runner1928 approach? Xaris333 (talk) 21:01, 12 December 2017 (UTC)
Perhaps you could use wears (P3828). Runner1928 (talk) 20:47, 12 December 2017 (UTC)
@Runner1928: You mean something like Real Madrid CF (Q8682) --> wears (P3828) --> Real Madrid C.F. home kit (Q28199035)? Xaris333 (talk) 20:52, 12 December 2017 (UTC)
Exactly. It looks like start time (P580) and end time (P582) would be appropriate qualifiers too. Runner1928 (talk) 20:55, 12 December 2017 (UTC)
@Runner1928: It may works. The only problem I can find is how to specify that Real Madrid C.F. home kit (Q28199035) is the home kit, Real Madrid C.F. away kit (Q45321396) the away kit and Real Madrid C.F. third kit (Q45321398) the third kit? Xaris333 (talk) 21:01, 12 December 2017 (UTC)
wears (P3828) uses astronaut (Q11631) space suit (Q223571) as one example, so I think using it for sport uniforms would also be okay. criterion used (P1013) could be used to specify home kit/away kit etc. ChristianKl (✉) 21:03, 12 December 2017 (UTC)
@ChristianKl: Ηοw? Xaris333 (talk) 21:14, 12 December 2017 (UTC)
https://test.wikidata.org/wiki/Q110290 ChristianKl (✉) 21:23, 12 December 2017 (UTC)


@Swpb, Dhx1, Pasleim: Do you agree with this approach? Xaris333 (talk) 21:28, 12 December 2017 (UTC)


I have completed Real Madrid C.F. home kit (Q28199035), Real Madrid C.F. away kit (Q45321396) and Real Madrid C.F. third kit (Q45321398) and added them to Real Madrid CF (Q8682). Please check that 4 items and make suggestion how we can improve them. Xaris333 (talk) 23:21, 12 December 2017 (UTC)

Instead of the statement instance of (P31)=association football kit (Q1381731) with qualifier applies to part (P518)=home kit (Q45321977), I would directly use instance of (P31)=home kit (Q45321977) --Pasleim (talk) 10:20, 13 December 2017 (UTC)
@Pasleim: home kit (Q45321977) is not only for football. So, we can't have subclass of (P279)=association football kit (Q1381731). Of course we can change the label to Home kit (association football). Xaris333 (talk) 11:49, 13 December 2017 (UTC)
I dislike having so much information in qualifiers for instance of (P31). I would add a new item "association football home kit" that subclasses both "home kit" (general) and association football kit (Q1381731). ChristianKl (✉) 12:46, 13 December 2017 (UTC)
We have a more specific property for use. use="home kit" should work.
--- Jura 14:06, 13 December 2017 (UTC)
On what item would you add uses (P2283) "home kit"? ChristianKl (✉) 15:20, 13 December 2017 (UTC)
It's has use (P366) ;)
--- Jura 17:13, 13 December 2017 (UTC)
@Jura1: Can you change the item to understand what you mean? Xaris333 (talk) 12:06, 14 December 2017 (UTC)
@Xaris333:   Done at Real Madrid C.F. home kit (Q28199035)
--- Jura 12:08, 14 December 2017 (UTC)
Thanks. Xaris333 (talk) 12:58, 14 December 2017 (UTC)

And one more question: do you prefer to have an item per period (like Real Madrid C.F. home kit (Q28199035)) or a general item "Real Madrid C.F. home kit" and make the changes every year? I am asking because an item "Real Madrid C.F. home kit 2016-17" maybe useless now. And would be easier for the user to change an exist item than create a new one. On the other hand, by creating each period a new item, we can have a "library" for all home kit of Real Madrid during the years. Xaris333 (talk) 16:05, 13 December 2017 (UTC)

It's my intuition to prefer are more general item that qualifies all the relevant statements with "start time"/"end time" and sets the most uptodate state as referred status would be the best but I don't have much subject matter experience, so I don't have a strong opinion. ChristianKl (✉) 16:42, 13 December 2017 (UTC)
So you mean, for example, to have many body (Q170494) values for has part(s) (P527) and have start time"/"end time with it? Xaris333 (talk) 12:06, 14 December 2017 (UTC)
Yes. ChristianKl (✉) 12:50, 14 December 2017 (UTC)
Works for me. Thanks. Xaris333 (talk) 12:58, 14 December 2017 (UTC)

Thank you all for your help. I will start apply it. I don't know any tool that can help me. I think I have to do it by hand. Again, thanks for you help. Xaris333 (talk) 12:58, 14 December 2017 (UTC)

Gathering examples of the search engine not ranking items well

I dislike that when I type "male" into sex or gender (P21) the item I search is on the third place. When I talked with Lydia, she recommended gathering more examples of where the search engine gives bad sorting. I created https://www.wikidata.org/wiki/User:ChristianKl/List_of_bad_sorting as a list to gather examples. If you have any interesting ones, I invite you to add them to the list. ChristianKl (✉) 23:20, 12 December 2017 (UTC)

  • As P21 has a "one-of" constraint, selection could easily suggested its values first. I think I once read a proposal for that. Didn't you write that?
    --- Jura 17:09, 13 December 2017 (UTC)
  • My first proposal was that constraints in general result in items that violate the constraints being listed after the ones that do. That seems to be computationally too costly.
Limiting this to the one-of constraint would be a possibility that's computationally tractable but it won't fix the issue that "male organsim" gets ranked before "male" because "male organism" has >10 sitelinks and "male" has none. It would get "male" only to the second place by letting it rank better than Malé (Q9347). ChristianKl (✉) 13:24, 14 December 2017 (UTC)

Query returning same item several times

As I am still new to writing queries I am probably missing something, but can someone explain me why this query returns several items more than once. Whether I write just SELECT or SELECT DISTINCT makes no difference. Ahoerstemeier (talk) 12:01, 14 December 2017 (UTC)

Here is a version that opens as list. It seems to be mainly because some parks have multiple coordinates and the query retrieves the same coordinates in two different ways (wdt:P625 and p:P625). Not sure why Q1322875 appears 4 times as well. It disappears if you retrieve just as wdt:P625 [11]
--- Jura 14:14, 14 December 2017 (UTC)

Is Q42086973 a sandbox?

Otherwise, the content in there does not make much sense to me. If it is, it should probably be marked a such, or merged. --Daniel Mietchen (talk) 00:48, 19 December 2017 (UTC)

It seems like the user who created it has only created this one item. Without knowing Chinese it's however hard to know what the item is supposed to be about. ChristianKl (✉) 00:51, 19 December 2017 (UTC)
@Daniel Mietchen, ChristianKl: The labels, when popped into Google Translate, say that the item's about the item's creator, but as the statements are pretty incoherent otherwise, I'd probably go ahead and delete it. Mahir256 (talk) 04:51, 19 December 2017 (UTC)
Now deleted; thanks for reporting. —MisterSynergy (talk) 09:56, 19 December 2017 (UTC)
This section was archived on a request by: Andy Mabbett (Pigsonthewing); Talk to Andy; Andy's edits 00:14, 21 December 2017 (UTC)

Rump = Rumpius

Daniel Rump (Daniel Rump (Q1162406)) is the same person as Daniel Rumpius (Daniel Rump (Q46242258)). --Gerda Arendt (talk) 19:04, 20 December 2017 (UTC)

It appears that someone has now redirected the latter to the former. Andy Mabbett (Pigsonthewing); Talk to Andy; Andy's edits 00:04, 21 December 2017 (UTC)
This section was archived on a request by: Andy Mabbett (Pigsonthewing); Talk to Andy; Andy's edits 00:13, 21 December 2017 (UTC)

IP edits

Do we have any research showing which percentage of IP edits is vandalism? We definitely have a lot of IP vandalism in highly visible articles, which stays there for hours and sometimes days, which basically compromises on the idea of using Wikidata directly on the projects. Do we know whether we at least have sufficient benefits from IP edits?--Ymblanter (User talk:Ymblanter) 13:58, 10 December 2017 (UTC)

No research, but some experience: The amount/percentage of vandalism strongly depends on topics and language. I had an eye on German labels and descriptions a while ago using Pasleim’s reCh-Tool and stopped doing so after barely any vandalism by IP and new users showed up. Tha vast majority of edits was good, or at least clearly no vandalism. Spanish and English labels and descriptions, however, seem to attract much more bad-faith edits. —MisterSynergy (talk) 06:31, 11 December 2017 (UTC)
I see on a regular basis vandalism in English and Russian on my watchlist.--Ymblanter (talk) 09:42, 11 December 2017 (UTC)
I've been patrolling recent changes an hour or so each week and there's definitely both good and bad from IP edits. Some IP edits do a lot and seem very conscientious, I think we would lose quite a bit in banning them all. There are also named users who do damage. I think the main thing we need is better organization of our patrolling. How do other wiki's handle it? ArthurPSmith (talk) 15:28, 11 December 2017 (UTC)
As you probably know, in the English Wikipedia there is no organized patrolling except for (i) new page patrol, which goes to the queue, and (ii) pending changes on the pages where those are installed, (iii) ClueBot. Pending changes, installed on the most visible pages, could actually help, I proposed this here some time ago but received no support. Some projects, for example, the Russian Wikipedia, use flagged revisions. All pages in the main space are subject to these flagged revisions, which are by default shown to the readers. Anybody with the flag can patrol some of all of the flagged revisions, and there is a queue organized. When I last edited there more than six years ago, the project was clearly understaffed, and the queue was steadily growing. Sometimes a user would decide to flag thousands revisions per day, which resulted in bad quality (borderline vandalism or incompetent) edits getting through; nobody cared. I do not think we have enough manpower for flagged revisions.--Ymblanter (talk) 15:44, 11 December 2017 (UTC)
Last time I checked the recent changes for properties for ip edits, it turned out to be 80%+ vandalism. We need a better way to deal with this vandalism otherwise we end up with blunt solution like semi-protecting all properties. Multichill (talk) 16:57, 11 December 2017 (UTC)
For properties, to be honest, I do not see why they are not yet all semiprotected. There are not so many of them, we can easily patrol all of them and answer requests (and the only type of requests I can imagine for properties would be adding translations), and vandalism in properties can do too much damage. However, I am wondering whether we should be better off if Ips could not edit items (semi-protection of all items is not an option since project editors who are not autoconfirmed on Wikidata must be able to move and delete pages.--Ymblanter (talk) 09:55, 12 December 2017 (UTC)
I would support semiprotecting all properties. Even when it comes to the translations it's preferable when they are set by people who understand Wikidata and have already made a bunch of other edits. ChristianKl (✉) 13:28, 12 December 2017 (UTC)
Should we start an RfC on this issue?--Ymblanter (talk) 14:55, 12 December 2017 (UTC)
We must have already had one, at least here at the project chat... Matěj Suchánek (talk) 15:05, 12 December 2017 (UTC)
We have too much vandalism on Wikidata, and it's a lot of work to check and find bad edits. It's also one reason why many Wikipedias don't want to use Wikidata in their wikis because then the vandalism made on Wikidata appears on Wikipedia. We had at least two previous discussions about using the Pending changes on Wikidata: Wikidata:Project_chat/Archive/2013/05#Pending_changes and Wikidata:Project_chat/Archive/2017/09#Pending_changes?. I think the extension would be interesting to see on Wikidata, but I think it would not work on Wikibase system as it's now. Users interested of Pending changes can also read m:Requests for comment/Flagged revisions deployment. I fear we don't have enough interested users to check the pending changes and the queue would get too long. It works very well on some wikis, for example on the Finnish Wikipedia the waiting time before edits get approves is usually very short. Stryn (talk) 17:15, 11 December 2017 (UTC)
One solution to getting better at fighting vandalism is to increase the amount of people who follow commonly used items, by letting people automatically follow the item towards which a statement they point as I laid out in my wishlist item.
The other is to get better integration into the Wikipedia watchlists so that vandalism that affects Wikipedia shows up on their watchlists.
I don't think pending changes would be a good step for Wikidata. ChristianKl (✉) 18:09, 11 December 2017 (UTC)
When it comes to Wikis that do use pending changes, there's no reason why they couldn't also use that mechanism for Wikidata related changes. ChristianKl (✉) 20:26, 11 December 2017 (UTC)
Great ideas by ChristianKl here! If the right tools existed, Wikidata would make patrolling more efficient for the Wikipedias, thanks to more people (from various Wikipedias) being available in many timezones to mark edits as patrolled. Syced (talk) 06:26, 12 December 2017 (UTC)
My experience is that the majority of Wikipedia editors will never consciously edit Wikidata (they still get Wikidata edits by moving pages for example, but they would never go and make edits on purpose). That watchlists currently suck does not exactly help.--Ymblanter (talk) 14:54, 12 December 2017 (UTC)
Hopefully that will change in the future when we will have a way for them to edit Wikidata directly from their infoboxes. ChristianKl (✉) 22:06, 12 December 2017 (UTC)
That will multiply our burden to fight this kind of "improvements". --Succu (talk) 22:13, 12 December 2017 (UTC)
The current trend in the English Wikipedia is such that most likely in a year, after a series of RfCs, it will be prohibited to display Wikidata directly in infoboxes in the main namespace. Wikidata vandalism (and our inability to deal with it efficiently) is the most significant reason for that.--Ymblanter (talk) 11:40, 13 December 2017 (UTC)
This is IMO a very pessimistic view on the enwiki activities these days. (Since I’ve read some of the discussions I understand how you came to that conclusion, but I don’t share it.) Yes, there are a couple of very loud editors who are aggressively lobbying against Wikidata, but I can’t see that there is a clear majority of enwiki editors in support of a full Wikidata ban in main namespace. Let’s see how it develops. —MisterSynergy (talk) 12:08, 13 December 2017 (UTC)
Indeed we need to see what happens when actual RFCs are up and who comes to argue what opinions. I however share the concern that Wikidata vandalism is a serious problems, and, in particular, when vandalism in highly visible items does not get reverted for hours and sometimes for days. We must have some solutions at least for the most visible items, even if banning IP edits is not a solution.--Ymblanter (talk) 15:17, 13 December 2017 (UTC)
My last idea on this problem is to have two dispatch queues (ie. mechanism that is used to update Wikidata content in infoboxes). The main one would be for trusted users, like bots, admins or confirmed users, which would work in real time, the second one would only dispatch edits from IPs and newbies when they get patrolled. (Or to only enqueue edits if they are patrolled but that's just implementation detail.) Matěj Suchánek (talk) 15:05, 12 December 2017 (UTC)
any idea of what fraction of edits that need patrolling actually get patrolled here right now? ArthurPSmith (talk) 19:33, 12 December 2017 (UTC)
@ArthurPSmith: You can compare some filters on the new, confusing Special:RecentChanges with the patrol log to get a feel for it. —Justin (koavf)❤T☮C☺M☯ 20:38, 12 December 2017 (UTC)
I don't understand users who call this new feature confusing. It brings new important features for patrolling we didn't have before. It's something Wikidata should take advantage of as soon as possible to combat vandalism more effectively. Matěj Suchánek (talk) 20:52, 12 December 2017 (UTC)
@Matěj Suchánek: Its interface is entirely unlike anything else in MediaWiki and how to add filters is counterintuitive to me. —Justin (koavf)❤T☮C☺M☯ 21:01, 12 December 2017 (UTC)
Last month, there were 33,000 non-confirmed edits, of which almost 6,000 have been patrolled. Matěj Suchánek (talk) 20:52, 12 December 2017 (UTC)
I don't suppose there's any well to tell how many of those have been reverted? Can we tell how many are no longer the latest revision? But that doesn't necessarily help (if one IP makes 10 edits only the last one would be latest, but all would still be active). ArthurPSmith (talk) 21:00, 12 December 2017 (UTC)
I updated the query, the results should be more accurate now. Matěj Suchánek (talk) 21:20, 12 December 2017 (UTC)
In that query what's the difference between "non current" and "non current patrolled"?ChristianKl (✉) 12:54, 13 December 2017 (UTC)
"non current patrolled" ⊆ "non current" (patrolled ∪ unpatrolled). Matěj Suchánek (talk) 17:29, 13 December 2017 (UTC)
  • Ok, thanks for those numbers. It looks like we're doing some of the work needed, but we need to add about 4x our current effort to cover it all. Looking at the manual patrol log it seems we have maybe half a dozen regular users doing it all, plus DeltaBot's major work in this area. @YMS: seems to do by far the most among regular users - perhaps YMS has some insights on how we can improve here? It seems like we at least need a couple dozen more people working on this to tackle it properly. ArthurPSmith (talk) 20:27, 13 December 2017 (UTC)
I don't think I have any particular insights to share, but I indeed use a specialised tool I wrote for myself. It sure won't be ideal for everybody; it won't even be the best choice RC tool for everybody willing to use an RC tool, but I believe it to be a very powerful tool and I hope that I will be able to release a beta version of it to the public by Christmas. This will happen at Wikidata talk:WikiProject Counter-Vandalism first, so if anyone wants to give a try to yet another RC tool, watch this. --YMS (talk) 22:01, 13 December 2017 (UTC)
Could we in the meanwhile adapt this tool to Wikidata? I was trying yesterday, but without much success, and it is very helpful to patrol multiple good edits in the same item.--Ymblanter (talk) 06:11, 15 December 2017 (UTC)

George Cornelius Gorham

I am attempting to link Q5538125 to wikisource:Author:George Cornelius Gorham, but the save button is greyed out. Would someone be able to give me a hand? Thanks, 142.161.81.20 21:30, 14 December 2017 (UTC)

Seems to be done special:diff/609122838.  — billinghurst sDrewth 00:00, 15 December 2017 (UTC)

Wikimedia Commons links being implemented

 
WikidataLovesCommons

There was a previous discussion here in the project chat last month where someone asked if WD:N be expanded to include Wikimedia Commons categories (see this) and there was an overwhelming support, as Wikimedia Commons itself is busy with structuring its data wouldn't it now be time to add a Wikimedia Commons link 🔗 option instead of directing users to “Other sites”? I agree that which Wikimedia Commons categories get linked should be heavily scrutinised, but it should be time to technically bridge 🌉 Wikidata with Wikimedia Commons.

Sent from my Microsoft Lumia 950 XL with Microsoft Windows 10 Mobile 📱. -- 徵國單  (討論 🀄) (方孔錢 💴) 09:41, 15 December 2017 (UTC)


I have Q45747062 a case where a Commons category (Category:Journal de Bruxelles nr 76) needs to be added. See also next subject.Smiley.toerist (talk) 14:26, 15 December 2017 (UTC)
  Done if you check how I did it, you can do it yourself for the other ones. Mbch331 (talk) 17:32, 15 December 2017 (UTC)

Joining wikidata, the Commons and French Wikimedia

'Journal de Bruxelles' was revolutionairy Belgian periodical. Two issues from the year 1799 have been digitalized in French Wikisource. It also exists as a Commons category. There is also a article in French Wikimedia about 'Journal de Bruxelles', only the wrong one, as there where several periodicals with the same name. Can this be added to Wikidata? Is Wikisource excluded from Wikidata? (As I see no Wikidata link in Wikisource) As the content contains news, these could be linked to certain subjects such as ships, generals, etc. The links are the following:

Journal de Bruxelles (1790-1800)/76-1799 https://fr.wikisource.org/wiki/Journal_de_Bruxelles_(1790-1800)/76-1799 https://fr.wikisource.org/wiki/Livre:Journal_de_Bruxelles_nr_76_1799 https://commons.wikimedia.org/wiki/Category:Journal_de_Bruxelles_nr_76

Journal de Bruxelles (1790-1800)/83-1799 https://fr.wikisource.org/wiki/Journal_de_Bruxelles_(1790-1800)/83-1799 https://fr.wikisource.org/wiki/Livre:Journal_de_Bruxelles_nr_83_1799 https://commons.wikimedia.org/wiki/Category:Journal_de_Bruxelles_nr_83

General article https://fr.wikipedia.org/wiki/Journal_de_Bruxelles, but as there are several periodicals named 'Journal de Bruxelles' this is the wrong one.Smiley.toerist (talk) 11:18, 15 December 2017 (UTC)


Linkedin

Property:P2035 was formatted as a full url, is there anyway we can change it to just the ID portion of the url and make it a true identifier? For Jared Robert Banta (Q45753535) we add in https://www.linkedin.com/in/jared-banta-7a369a48. It would be better if we had it as "jared-banta-7a369a48" and made it a true identifier? Any thoughts? --RAN (talk) 17:27, 15 December 2017 (UTC)

  • They used to have a different format, incompatible with that datatype. As we attempt to provide stable properties, I don't think it's good to shuffle it around. On the mere GUI side, I think it could be displayed with others further below.
    --- Jura 17:44, 15 December 2017 (UTC)

Categories for alumni

Hoi, I have been active again on categories for alumni. I did it in the way I used to and consequently I used "is a list of" instead of "category contains". I do not know how to change this and retain the qualifier.. How to do this? Thanks, GerardM (talk) 11:00, 21 December 2017 (UTC)

So mostly these ones, I guess, and they should be moved from is a list of (P360) to category contains (P4224)? —MisterSynergy (talk) 11:20, 21 December 2017 (UTC)
Yes all of them. Thanks, GerardM (talk) 11:39, 21 December 2017 (UTC)
@GerardM: all done. —MisterSynergy (talk) 21:14, 21 December 2017 (UTC)
This section was archived on a request by: MisterSynergy (talk) 21:14, 21 December 2017 (UTC)

Beatrice von Dovsky (Q88821)

Beatrice von Dovsky: birth and death month got mixed up; aka Beatrice Dovsky, Beatrice Drewikowsky (birth name, also spelled Drevikowsky), Beatrice von Vay (married name), see German Wikipedia for sources or look here pdf p. 955. -- 84.41.34.154 09:28, 10 December 2017 (UTC)

  • I added the dates found on the gravestone. Depending on what is thought to be preferred or incorrect, ranks can be set.
    --- Jura 09:51, 10 December 2017 (UTC)
    • GND doesn't list any months. It was a mix-up, pure and simple. Please delete the wrong dates and please also add the akas. -- 84.41.34.154 10:16, 10 December 2017 (UTC)
      • GND generally has the full dates in the rdf or some other format. As such the other dates shouldn't be deleted. If they are incorrect, they can be set to "deprecated rank". As the item isn't protected, you can change the ranks and add more references directly.
        --- Jura 10:42, 10 December 2017 (UTC)

Help:Sources doesn't include gravestones

What would be a good way to include them? At Q88821#P569, there is now type of reference (P3865)=tombstone (Q203443) and the image (P18) with the image at Commons. A bit basic, but it could be a start.
--- Jura 22:07, 15 December 2017 (UTC)

Facto Post – Issue 7

The latest issue of Facto Post is available via this link. Charles Matthews (talk) 13:28, 16 December 2017 (UTC)

UK places with multiple heritage listings

I'm having another issue with UK listed buildings (so paging @Anvilaquarius: who came in on the last such thread).

Background: I have been preparing an upload to Commons of a substantial number of old maps and plans from 19th century books. (See batch preparation pages -- UK initially, but much more from the rest of the world to follow).

Through a process of georeferencing, I have the coordinates of the corners of the map or plan. Given these co-ordinates I have been using Wikidata to help identify what the map or plan may actually be of, for features up to about 250m from side to side. (For bigger objects than this, there are generally lots of Wikidata location hits, and it's difficult to filter which ones might be most likely for somebody to have made a map of, of that size -- but if people have some good ideas I would be very interested. Identification is already fairly unreliable at the 200m scale).

The identification process I have been using has been quite simple minded:

  • Find the nearest Wikidata item to the centre of the map, that has a Commons category (P373); or, failing that
  • Find the nearest Wikidata item.

Using the Wikidata look-up I have also been able to separate out plans for cathedrals, castles and churches / abbeys / monastic ruins.

But I have hit a problem with some places for which we have multiple items -- for example, looking at the Commons category page for c:Category:Gloucester Cathedral, it has four entries in the register of listed buildings, each of which we have a different item for, all of which link to the Commons cat, but only one of which (and not the nearest to the centre of the map) is instance of (P31) cathedral (Q2977). Similarly, here's the Commons category for c:Category:Guildford_Castle, with 5 different listings.

What should be done in such cases?

My initial reaction was to think the items should be merged here. But now I think (with a few exceptions) that that is probably not the right solution. The exceptions I would make, that IMO should be merged would be eg two ends of the same bridge, that happened to be different local authority areas; or places that have both a 'listed building' and a 'scheduled ancient monument' listing. Where these refer to the same place, I think it does make sense to combine them. But for eg Gloucester Cathedral, perhaps it does make sense to have separate items for eg the Cloisters and the Chapter House, since these are distinct identifiable structures within the whole. Perhaps then Cathedral Chapter House (Q17525927) and Cathedral Cloister And Lavatorium (Q17525928) should indeed remain separate from Gloucester Cathedral (Q262500), perhaps linked by part of (P361), with Gloucester Cathedral (Q262500) having the Gloucester Cathedral commonscat link. I'm less sure about Gloucester Cathedral (Q17525925), which maybe should be merged. Similarly, for Guildford Castle, does it make sense to keep all five separate listed items, such as Remains of Shell Keep at Guildford Castle (Q17527536), marking them all as part of (P361) Guildford Castle (Q139629), with only the latter getting the Commons category (P373) ?

On the other hand, that would mean that we were divorcing the item with the Wikipedia link and the Commons cat from its historic buildings listings. Is that a good idea? And does this make for too much of a challenge for people to know when to merge items and when not to merge them?

What thoughts do people have? Jheald (talk) 18:18, 13 December 2017 (UTC)

Discussion also noted at Wikidata_talk:WikiProject_Built_heritage#Places_with_multiple_heritage_listings. Feel free to move the thread there if people think it would be more appropriate. Jheald (talk) 18:30, 13 December 2017 (UTC)

I do not quite understand what the problem is. Using the example of the Gloucester sathedral, why can not one have one item about the cathedral (with sitelinks to Wikipedia articles) and five items which list parts of the cathedral which are listed buildings (referenced properly) and possibly having separate Commons categories?--Ymblanter (talk) 18:32, 13 December 2017 (UTC)
I see it exactly like Ymblanter above. Reality is not as hierarchical as we would like it to be. Hence, listed buildings and building parts will alawys be different from the scope of Wikipedia articles and from the scope of Commons Categories, or from the scope of authority data entries. There's not much to be done about it but to try to describe the items as understandable as possible and link them via "part of" and similar properties. --Anvilaquarius (talk) 18:39, 13 December 2017 (UTC)
There are a couple of lighthouses with multiple listings as well. I linked them together with "has part" and that seems to work fine. Looking for images for the items, I found that Commons could use separate category for the foghorn of one of them ..
--- Jura 18:47, 13 December 2017 (UTC)

Thanks, everybody. Here's a query for multiple article-items pointing to a single Commons category, with at least one item having a National Heritage List for England number (P1216): tinyurl.com/ybdakrac. 1275 rows, so there may be some cleaning up to do. Jheald (talk) 22:51, 13 December 2017 (UTC)

Same building, different name / different use?

Next question: what about when we have the same building complex, re-purposed to have a slightly different use, under a slightly different name: eg

Gloucester Abbey (Q5572012) -> Gloucester Cathedral (Q262500) in 1540.

Is the first a facet of (P1269) of the second? Was it followed by (P156) the second? Property structure replaced by (P167) is one that is not applicable, I think, because the structure was not knocked down and replaced.

Also, what is the appropriate inception (P571) for Gloucester Cathedral (Q262500) ? The date of the first religious structure on the site? The date of the oldest section of the current building? The date on which it was re-purposed to its current use? All of the above, with suitable qualifiers, eg determination method (P459) -> a value representing one of the above, identity of subject in context (P4649) -> (eg) Gloucester Abbey (Q5572012) ? Jheald (talk) 00:32, 14 December 2017 (UTC)

@Jheald: It is always a definition problem: if I take the definition of Gloucester Abbey (Q5572012), this is an abbey (Q160742) which is a special case of monastery (Q44613). What is a monastery (Q44613) ? This an instance of institutional complex, place of worship and organization.
So if we take the definition of institutional complex for monastery, then Gloucester Cathedral (Q262500) is part of Gloucester Abbey (Q5572012), as the church is only a building among others in a monastery. Then for the inception date, the inception date for Gloucester Abbey (Q5572012) (church, monks settlement,...) is the date when the first building was created and for Gloucester Cathedral (Q262500) this is the date when the current building was erected. Snipre (talk) 15:57, 14 December 2017 (UTC)
The WP:en article is not clear as it defines the abbey as being the church, which not true. Snipre (talk) 16:09, 14 December 2017 (UTC)
From my understanding, Gloucester Cathedral (Q262500) was a abbey church (Q334383) between 1058 and 1540, and since 1540 it is a cathedral (Q2977). Snipre (talk) 16:16, 14 December 2017 (UTC)
Thanks @Snipre:. So you would recommend Q262500 be considered an item for the building(s) rather than the institution, and therefore perhaps have:
In general terms, in English I think both the words "cathedral" and "abbey" probably both encompass the entire complex including cloisters, chapter-house, etc; rather than just the church building specifically. So perhaps P31 "abbey" rather than "abbey church" for the earlier period.
What would you recommend for Gloucester Abbey (Q5572012), to indicate its relation to Gloucester Cathedral (Q262500) and its subsequent re-use as a cathedral? Jheald (talk) 17:50, 14 December 2017 (UTC)
@Jheald: I agree with your 2 first points but not with the reuse of inception (P571): the building was erected once, not twice. Use of inception (P571) to describe status change is wrong for me: better use significanr event property with a appropriate event to define the status change. Snipre (talk) 23:41, 14 December 2017 (UTC)
The question is, though, do we use Q262500 to represent "the" building, or the whole sequence of building complexes and associated institutions.
Similarly, should St Paul's Cathedral (Q173882) just be the building designed by Wren, or the institution of a cathedral dedicated to St Paul from AD 604 onwards?
In some ways it makes sense to have a primary item that is as general and encompassing as possible, in both space and time, and then further items that may concretely represent particular parts of the building or specifical constructions in its history, such as Old St Paul's Cathedral (Q2576524).
Or should we keep Q173882 for just the present-day structure? Jheald (talk) 00:34, 15 December 2017 (UTC)
In some ways there are parallels with the question of "my grandfather's axe" (or alternatively the Ship of Theseus (Q1050837)). Can my grandfather's axe still get a single item, if the head has been replaced twice and the handle three times? Jheald (talk) 00:43, 15 December 2017 (UTC)
@Jheald: Again this is the definition of the item which should answer your question, not me: a building can't be an association so you can't mix both concepts in the same item. Then you have to keep a logical structure so if you say that Gloucester Cathedral (Q262500) was once the abbey church of Gloucester Abbey (Q5572012) then you can't consider now the other buildings of the abbey as being elements of the Gloucester Cathedral (Q262500) defined as cathedral. But this las point is more a personal point of view. And for your last question about your "my grandfather's axe", I proposed you to look at any items about a person: does the item refer to the person as a baby, teenager or adult ? Was Albert Einstein (Q937) a physician since he was born or only after he finished his studies ? Snipre (talk) 23:05, 16 December 2017 (UTC)
  • It's hard to say. It seems that initially the cathedral was the church of the abbey. Now it seems to be the opposite: abbey buildings are an annex of the cathedral.
    --- Jura 16:10, 14 December 2017 (UTC)

Two commons links

We have Albert Memorial (Q281465) and Category:Albert Memorial (Q24704063), apparently just so that we can link to commons:Albert Memorial and commons:Category:Albert Memorial (respectively). Andy Mabbett (Pigsonthewing); Talk to Andy; Andy's edits 20:50, 14 December 2017 (UTC)

What a mess. The two items need to be merged, and the merged item should have both a Commons category (P373) property and a Commons gallery (P935) property. As to whether the category or the gallery gets the sitelink, that's a coin-toss, but I would sitelink the category. Jheald (talk) 00:46, 15 December 2017 (UTC)
The whole system needs to be corrected. Galleries on the commons are almost always just irrelevant. The most important thing about the Commons sitelink is to get from the main Commons entry about an item to Wikidata quickly (and to quickly see if such an entry exists at all), so links from a crappy gallery page that no one ever visits are pretty useless. Jheald's proposal seems sensible, but only if Commons finally starts to display links to Wikidata derived from Commons Category and Commons Gallery property entries, not only from the flawed sitelinks entry. --Anvilaquarius (talk) 09:18, 15 December 2017 (UTC)
For the sitelinks, an alternative would be to create an item "Category:Albert Memorial", connected to Albert Memorial (Q281465) by category's main topic (P301)/topic's main category (P910) in each direction, sitelink that new item to the category, and put c:Template:Interwiki from Wikidata on the top of the Commons category. Both the category-item and the article-item here would also have Commons category (P373) statements. That method is probably how to maximise the interwiki linking from Commons to Wikipedias.
@Anvilaquarius: It doesn't help the multitude of casual browsers of Commons, but if you're doing a lot of work on Commons and Wikidata, you may find the wdcat.js script helpful, which shows when a Commons category is the object of an incoming Commons category (P373) statement on Wikidata. The script gets its information from a SPARQL lookup, so probably can't be scaled for mass usage, which for the moment has to rely on sitelinks and templates, but as an individual user I find it quite useful. Jheald (talk) 10:43, 15 December 2017 (UTC)
User:Matěj Suchánek has gone for a different solution, which is to mark Q24704063 as a Wikimedia permanent duplicate item (Q21286738). But since the Commons interwiki template can't interpret that, and there is no language that has a Wikipedia alternate article, I'm going to change it to instance of (P31) Wikimedia category (Q4167836), as discussed above. Jheald (talk) 12:09, 15 December 2017 (UTC)

Q44280387

Is this a property that created wrongly in the Main namespace? --Liuxinyu970226 (talk) 01:41, 17 December 2017 (UTC)

OpenStreetMap node (Q42375175) seems likely. --Liuxinyu970226 (talk) 01:48, 17 December 2017 (UTC)
I ping Gnjo55 and Martinbayo to know their opinion. Pamputt (talk) 10:38, 17 December 2017 (UTC)

Linking to Draft namespace

Tracked in Phabricator
Task T183100

Hello. The topic was raised on the talk page of the Everipedia (Q44960346) item, which is linking to the Draft namespace of the English Wikipedia. It seems difficult to realize that you land in a draft when you come from another Wikipedia following the interwiki link. Have there been any previous discussion on this topic? — Envlh (talk) 19:32, 14 December 2017 (UTC)

I believe we used to not link at all to the Draft namespace, but then ran into problems when people moved pages from Main to Draft and back again (wikidata links, and therefore interlanguage links, were lost). Not sure what the policy is on this here at the moment. ArthurPSmith (talk) 19:38, 14 December 2017 (UTC)
Links to the draft namespace shouldn't exist, and in previous cases the English Wikipedia community was unhappy with them being set. I don't think we have to give EnWiki another reason to dislike Wikidata. Feel free to remove them when you see them. ChristianKl (✉) 19:39, 14 December 2017 (UTC)
Last discussion on this topic was in 2014, Wikidata:Project chat/Archive/2014/05#What do do with Wikipedia:Draft articles. Thoasp (talk) 19:42, 14 December 2017 (UTC)
I'm pretty certain that this wasn't the last discussion. ChristianKl (✉) 20:11, 14 December 2017 (UTC)
So Wikidata:Project chat/Archive/2017/02#Wikidata links to English Wikipedia draftspace? Matěj Suchánek (talk) 20:20, 14 December 2017 (UTC)
That seems more like the most recent one. ChristianKl (✉) 22:59, 14 December 2017 (UTC)
Empty for a few minutes, then this. --Larske (talk) 16:07, 15 December 2017 (UTC)

I suggest that we create an abuse filter to restrict adding the draft and other non-notable namespace links on Wikidata items. Stryn (talk) 16:31, 15 December 2017 (UTC)

I   Support such an abuse filter for draft. Which other non-notable namespaces do you have in mind? ChristianKl (✉) 19:28, 15 December 2017 (UTC)
Wikidata:Notability: "To be valid, a link must not be a talk page, MediaWiki page, special page, file, translations page, pages in User or Draft namespace, pages used by LiquidThreads (i.e. pages in Thread and Summary namespace), pages used by Structured Discussions." Stryn (talk) 20:39, 15 December 2017 (UTC)
When it comes to the user/draft namespaces I can easily see how you filter for them and that the filter would be good. In general an item that's notable might have valid links to pages that alone wouldn't grant it notability. Before introducing filters I would like to see for each filter a petscan search that shows what items currently link to that namespace. ChristianKl (✉) 01:01, 16 December 2017 (UTC)
We have had filters for user pages, template subpages and files for four years. Given how common draft articles are becoming, I can quickly make up a new filter for drafts. But I haven't seen much sitelinks to other forbidden namespaces, if any. Matěj Suchánek (talk) 09:44, 16 December 2017 (UTC)
I think such a draft filter would be good in the eternal project of removing reasons for EnWiki people to get annoyed at Wikidata. ChristianKl (✉) 13:20, 16 December 2017 (UTC)
I believe we can restrict links to specific namespaces in the config. I think that would be better than an abuse filter for performance reasons. Anyone up for creating a ticket? --Lydia Pintscher (WMDE) (talk) 18:02, 17 December 2017 (UTC)
  phab:T183100  — billinghurst sDrewth 20:53, 17 December 2017 (UTC)

Searching in phabricator

I was wondering how is the experience of other non-developers when searching in Phabricator? Can you usually find stuff? Have you tried searching for specific bugs? Personally, I haven't done so recently, but it happened that I couldn't even find requests I wrote there myself. I'm asking as some seem to want to use it for more Wikidata editing related stuff.
--- Jura 16:00, 15 December 2017 (UTC)

I can quite easily find the tasks what I want to find. https://phabricator.wikimedia.org/search/query/X10Ih0v0dkQo/#R tells you haven't started any tasks on Phabricator with account "Jura1". Stryn (talk) 16:37, 15 December 2017 (UTC)
@Stryn: This is in part because he goes by Esc3300 on Phabricator. Mahir256 (talk) 23:54, 15 December 2017 (UTC)
So it's doesn't work ;)
Can you find tasks about property suggestions? For some reason keyword search didn't always work for me.
--- Jura 16:49, 15 December 2017 (UTC)
Does this help? https://phabricator.wikimedia.org/search/query/.jWJkAE34cCe/ - PKM (talk) 19:23, 15 December 2017 (UTC)
  • Well, it's a general question about the experience people may have with phabricator for finding things (beyond clicking on a ticket number displayed in a thread).
    --- Jura 20:09, 15 December 2017 (UTC)
You're worried about using it more for editing related things. I think this can be helped by having specific boards for these specific tasks that are easier to scan than the monster Wikidata board. --Lydia Pintscher (WMDE) (talk) 18:05, 17 December 2017 (UTC)

Structured descriptions

Have WMF/developers ever thought of using structured Wiktionary items (when they exist), or internal Wikidata items, to generate Wikidata descriptions, instead of having the same generic "American actress"/"Wikimedia template"-type descriptions added and (mis-)translated manually over millions of items? (There is currently an English Wikipedia RfC related to this topic, which I have not participated in.) Jc86035 (talk) 14:58, 17 December 2017 (UTC)

This is the community board, for WMF/devs you’d have to ask at Wikidata:Contact the development team. Anyway, what is the problem with the current setting? Wikidata descriptions work well to my opinion and there is no need for changes. —MisterSynergy (talk) 15:31, 17 December 2017 (UTC)
What to do with Wikitonary items is a debate we can have in a year. ChristianKl (✉) 15:37, 17 December 2017 (UTC)
@MisterSynergy: If they worked well enough the WMF wouldn't have needed to disable them on the English Wikipedia.
Aside from the multilingual vandalism/test edits (difficult to find and tedious to correct), descriptions are mostly translated manually and so they may never get added/updated or are mistranslated, descriptions have to be added and updated manually for each of e.g. the three different varieties of English and the more than ten different variations of Chinese on every item, and frequently users just copy the name into the description for some inexplicable reason. There are probably other issues.
It would also be much more efficient, for example, to have a bot add two lexemes(?) which mean "Wikimedia template" to every item for a template, and have real people add translations to the lexemes "Wikimedia" and "template", instead of the current situation where bots go around slowly adding these translations to every single item. Similarly for every combination of demonymgender–occupationgender and so on (assuming there would be some sort of internal mechanism to indicate grammatical rules for word order etc. for each language).
There's not much point in asking the development team if no one else thinks it's a good idea, and quite often WMF employees will read and comment on this page anyway.
@ChristianKl: There's no harm in thinking about it now, especially to ask if this sort of thing (based on Wiktionary items, not Wikidata text fields) is feasible since it would probably require a sort of rudimentary translation mechanism for at least adjectives and nouns (assuming lexemes shouldn't be created for every combination of descriptions, anyway). Jc86035 (talk) 15:52, 17 December 2017 (UTC)
  • Descriptions were originally there to disambiguate items with identical labels in Wikidata. It seems counterproductive to do this with a highly unifying approach such as the one suggested by you. In some cases it would even be impossible, thus I like the flexibility of individual descriptions even though it is somewhat expensive to manage them.
  • It was much later that someone (at WMF?) realized that Wikidata descriptions are useful for other purposes as well, e.g. to disambiguate search results in mobile apps and so on.
  • The enwiki community fell for a couple of aggressive anti-Wikidata lobbyists, and now they have maneuvered themselves into a very difficult situation with only bad solutions left for implementation. IMO their decision does not really have anything to do with description quality or style.
  • Some tools provide automatic descriptions based on statements in the item (reasonator, User:Yair_rand/WikidataInfo.js, and so on). These descriptions do indeed describe the entities, but they are not at all unique in combination with the label and can’t fully replace the “description” field.
—MisterSynergy (talk) 16:08, 17 December 2017 (UTC)
en:MediaWiki:Wdsearch-autodesc.js can't replace the field in every case but at the moment for persons it almost always outplays fixed descriptions. The bigger part of the fixed descriptions also come from bots in a messy way (regarding dates of birth and death and thus hardly improvable by bot again. I think descriptions could benefit from a more systematic proceeding instead of the uncoordinated lonely bot operator actions. They are better than nothing, but much worse than en:MediaWiki:Wdsearch-autodesc.js and this isn't necessary. --Marsupium (talk) 16:32, 17 December 2017 (UTC)
One of the core ideas of these descriptions being shown in the App is that someone at the WMF thought that mobile readers who might not want to edit long Wikipedia pages might be willing to be motivated to add short descriptions texts. This was an idea to turn people who don't edit at the moment into editors who would hopefully edit more as they see how great it is to edit.
Wikidata building a fancy system of automatically created descriptions wouldn't help with that strategic goal. Better descriptions would be nice, but I don't see the strategic value in spending a lot of Wikidata development resources on it. ChristianKl (✉) 02:43, 18 December 2017 (UTC)

Consider merging Q19849270 and Q6867899?

The only conflict is in Cantonese Wikipedia, which IMHO we can encourage those users to discuss if both templates are mergeable or not. --117.14.250.110 02:23, 12 December 2017 (UTC)

Template:Bulletin (Q19849270) and Template:Centralized discussion (Q6867899) --Tagishsimon (talk) 02:42, 12 December 2017 (UTC)
Interesting, merging two templates are having much more possible than two single articles, hence pinging @Deryck Chan: in your opinion should we do that? Affected templates: yue:Template:Bulletin and yue:Template:Cent. --Liuxinyu970226 (talk) 10:52, 12 December 2017 (UTC)
Because Cantonese Wikipedia has a great tradition of following both English Wikipedia and Chinese Wikipedia practices unless they conflict or we create our bespoke system, we've adopted both templates for different purposes.
Normally I'd say   Disagree because these just aren't the same template, but since yue:Template:Cent is only used by a small set of pages, we have two   Split support and oppose options:
  1. Deprecate yue:Template:Cent and keep Template:Centralized discussion (Q6867899) only, knowing that this item will encompass a bunch of noticeboards that have different design intents.
  2. Keep the two items separate, move all templates based on zh:Template:Bulletin to Template:Bulletin (Q19849270), and said to be the same as (P460) or related property (P1659) the two items. Deryck Chan (talk) 15:15, 12 December 2017 (UTC)
  • This is not the place to discuss what yuewiki ought to do or not. We don't want to import debates from there here.
    Whether Wikidata items can be merged is determined by Help:Merge#Check_to_be_sure. I don't see how the two template items could satisfy that. I don't think people need to read Chinese to understand that.
    Possibly Template:Centralized discussion (Q6867899) needs to be split, but this isn't the question that was asked.
    --- Jura 15:31, 12 December 2017 (UTC)
    • @Jura1: I'm (putting my yue.wp admin hat on) offering to deprecate the little-used yue.wp template as one possibility of resolving this interwiki conflict, if we think that's the right thing to do despite Cent and Bulletin being two very different templates used by different Wikipedias to achieve similar purposes. Deryck Chan (talk) 15:23, 18 December 2017 (UTC)

Editing name, description, "also known as"

Wikidata seems recently to have unilaterally decided that when I go to edit name, description, & "also known as" that I'm only editing the English. This is annoying, given that I read/write several languages. I did not deliberately change any settings. Is this some sort of global change? Is there anything I can do to be able to get back to editing these fields in several languages? - Jmabel (talk) 07:01, 13 December 2017 (UTC)

You mean you can see no other language field than English? Beneath the table, there should be "All entered languages". Can you see that? Few weeks ago, several users reported a similar problem, which was then solved by null-editing their userpages. I have just done this for you. Matěj Suchánek (talk) 08:26, 13 December 2017 (UTC)
Didn't work at first, but seems to be working now. For what it's worth, I still had to play with it a little, but now it works as it should. I'm afraid I can't fully describe what "playing with it" consisted of. - Jmabel (talk) 16:09, 13 December 2017 (UTC)

{{Section resolved|Jmabel (talk) 16:09, 13 December 2017 (UTC)}}

  • Couldn't we add a note about this on top of the special page?
    --- Jura 17:11, 13 December 2017 (UTC)
    • @Jmabel: If you add a Babel tag to your user page to state all the languages you want to use (see my user page for an example), you can get all those languages displayed by default when you visit an item or a property page. For example I automatically see label boxes for Cantonese, English, Chinese, Japanese and German. What we seem not to have fixed yet is why these instructions aren't publicized on the interface. Deryck Chan (talk) 15:26, 18 December 2017 (UTC)
      • It wasn't a matter of what I see when I visit the page (got all the languages I normally use), it was that only one was appearing when I went to edit. And, as I say, it seems to be working now. - Jmabel (talk) 15:53, 18 December 2017 (UTC)

Query wikidata start getting messy because of "bad imports" of dates with different precisions and same rank

When more and more data is added I can see that we get a problem with dates added with different precisions and same ranks ==> you get a non "expected" result when asking queries

Problem creating queries gets very complex if you just would like to have back "one item"

  1. Suggestions:
    1. change the wdt prefix to just return the value with best precision if more has the same rank
    2. train people doing database imports of the consequences of what they do
    3. Histropedia added a new option "Merge duplicated results" link ?????


Example Lars Johan Werle (Q4107873) has more death dates P570 added with different precisions but same rank ==> we get 2 results back

The following query uses these:

  • Properties: family name (P734)     , date of death (P570)     
    SELECT ?item ?itemLabel ?death WHERE {
  ?item wdt:P734 wd:Q37507741.
  ?item wdt:P570 ?death.
  SERVICE wikibase:label {
    bd:serviceParam wikibase:language "en","sv" . 
  }
}
Try it!
- Salgo60 (talk) 09:15, 16 December 2017 (UTC)
  • If it's a problem at Hist..pedia, you might want to report it there.
    I don't see anything "bad" about the two dates of different precision and referencing, but with equal rank at Q4107873#P570. The next step could be that you add additional references for the more precise dates. Once done, you could set that date to preferred rank.
    --- Jura 09:27, 16 December 2017 (UTC)
No the Histropedia example was just one way of solving it. I guess they have seen the same problem
  1. Mass importing means +10 000 items added then people just jump to the next import it's feels like "touch and go"
    1. they never use the data ==> they dont see the problems....
    2. Example
      1. in Sweden for e.g. we have +7000 items already added with day as precision and rank normal using P3217 as reference ==>
        1. its research done by the Swedish National Archives
        2. its 4 full time very skilled historians quality assuring the result done by 50 professional authors everyone in the field knows about the quality work they do and this organization has done this since 1917 ==>
          1. uploading data with a date with less precision and same rank is wrong 99 times out of 100...
          2. one problem with Wikidata's designed today is that is not easy to understand the quality of P3217
        3. then someone upload another source with less good precision
        4. then all queries with wdt will return 2 results back. Yes you could better queries with grouping but querying Wikidata is already complex dont make it a "science...."
I suggested also a preview mode that could be used before mass import to see what constraints problems you will have see Talk Property constraints portal
- Salgo60 (talk) 10:00, 16 December 2017 (UTC)
  • I still don't get what's should be wrong with the two statements with equal ranks, but different precision/referencing at Q4107873#P570. It's part of Wikidata's design that multiple statements about the same fact can be added. It might even be the key improvement over some Wikipedias.
    --- Jura 10:08, 16 December 2017 (UTC)
You need to differentiate between
  1. how things are stored - its wrong to store two values with the same rank for P570 we have single value property constraints on P570
  2. how to query data - you use wdt to take the truth.... ==> for a date type we get a problem as we can have more values with same rank and different precision --> we have a problem
  3. expected result from the user querying the data - I guess normal is one record per person and his/her death date is expected when using wdt
The simple solution is to ask people to take more responsibility and check before uploading e.g. using Constraint Checker you violate Single_value and get the following "error"

 

The reality is that this check is not done by people batch uploading plus maybe also has less good understanding of the quality of data sources used.... houston houston we have a problem
With my query example above using wdt returns all values with same rank I expected one result row which was what I got before mass uploads ==> I can't trust my queries....
Solution: ??? How to handle uploads violating constraints rules??? The last months I have spent maybe 30 hours cleaning dats to avoid this error when people query...
-Salgo60 (talk) 12:09, 16 December 2017 (UTC)
  • The single value constraint on P570 isn't mandatory. Even if it was, it wouldn't override the general principle that multiple different statements about the same fact should be recorded. The solution is that either you need to evaluate the different sources and determine which one should get preferred rank or refine your query.
    --- Jura 12:39, 16 December 2017 (UTC)
    And one of my suggestions is that wdt should just return the result with best rank and best precision not more values....
    evaluate the different sources and do cleaning after other people before every query returning 10000 records ?!?!? Serious? - Salgo60 (talk) 12:54, 16 December 2017 (UTC)
    (edit conflict) Best precision does not imply best rank. There may be a date with precision to days but may be totally wrong. This is where ranks come to the scene. Matěj Suchánek (talk) 13:00, 16 December 2017 (UTC)
  • I have no problem understanding that I also can change all queries for me this is an "architectural problem" having prefixes like wdt you try to implement a concept that is easier for the person writing the query ==> returned is just the best ranked value. With the datatype "datetime" combined with "mass import" of sources with different precision we can see that we dont get just one value instead we get one value for every precision added..... In my book of design this is an user case not taken care of that will have a ”high" cost for people to write correct queries. My example with Histropedia was to point out that others have maybe already seen this problem and try to solve it in the query of Histropedia interface. For me that is the wrong place to correct it... User:ChristianKl I think points in the right direction to make it easier to write correct queries. Then the people doing mass import also need better feedback before starting a import that violates a lot of rules... Wikidata is growing and fixing everything by asking the users to write "better" more advanced queries will end in a "product" less usable - Salgo60 (talk) 16:24, 16 December 2017 (UTC)
  • There's no rule that it's mandatory to add a rank. Nonmandatory constraints suggests that it would be better if a person would judge which item should get preferred status but it doesn't say that a person who enters data violates our rules when they don't tackle the task of ranking. ChristianKl (✉) 17:00, 16 December 2017 (UTC)
@ChristianKl: I think this is another flavour of the problem
  • that we dont "rank"/"define quality" of sources.
  • that we dont communicate to the Wikipedia reader if the fact s based on an authoritative source or just a community created source
Lesson learned is most people in Wikidata dont have the skills/knowledge to understand every source quality e.g. I requested a year ago Property P3217 that is from the Swedish National archive(about SBL). They have 4 fulltime people just quality assure the 100 biographies they produce every year with the help of about 50 authors. When speaking to University teachers in history they tell me if students havnt checked with SBL they have to go back and do the homework....
In Wikidata/Wikipedia most people have no understanding of this quality ==> you start upload birth dates with precision year and "degenerate" Wikidata and they get no warning that we already have a value from a high quality source.....
I am convinced that both for people importing data and people reading Wikipedia with data "transferred" from Wikidata should get some indication of the quality and correctness of the information. On the session State of the project I asked that question (see 42 min video) and the answer was Trust is important and that is nothing yu get for free and it mus also be communicated to the readers of Wikipedia that this is data you can trust and not that if you are lucky no one has destroyed the quality by importing some lower quality data... - Salgo60 (talk) 20:17, 16 December 2017 (UTC)
You confuse multiple different issues:
(A) We show multiple values with wdt when we have multiple values stored. As far as trust goes, offering multiple values when we don't know which value is the best one is good. A reader who cares about which value is correct can look at the values we have with the corresponding sources and make his own judgement about which source is best. It's also great when he shares his research by setting appropriate ranks. Above you recommend that only one values is shown even when we have no reason to believe that it's better than the other values we store.
(B) The extend to which data from the Swedish National Archive should get preferred status. If you believe that it generally should get preferred status, run a bot that gives all the entries that are soruces from it preferred status. Of course, at best you would also check for cases where that means that there are claims with preferred to see what's up with those cases.
(C) When reading Wikipedia a reader isn't supposed to trust everything but check sources. Normally, there's no requirement in Wikipedia to source every claim. When Wikipedia imports Wikidata every imported claim has a source that a reader can judge for themselves and that's an improvement over the status quo of Wikipedia.
(D) It would be great to have additional ways to judge source quality. Signed statements will be one way. We might also have other ways in the future to automatically judge source quality but that isn't an easy task. ChristianKl (✉) 21:08, 16 December 2017 (UTC)
Lars Johan Werle (Q4107873) is a nice example. If you browse it you see that 23 June 1926 currently only has a week reference from Russian Wikipedia and three good references for 1926. The Primary source tool provides a link to a stronger reference. Encouraging the browsing of the item to actually look at the reference allows the data users to see the Primary source reference and add it (I did that in this case). Giving to the well developed Primary Sources tool it now has three entries ;)
When doing a mass upload, adding data as "normal rank" is correct behavior. ChristianKl (✉) 13:02, 16 December 2017 (UTC)
@ChristianKl: maybe time to redesign what correct behavior is.....Someone has to pay the price and right now its user querying who cant trust the result which is a very high price = in the end can we trust Wikidata?
Larske (talk • contribs • logs) helped me with a query ==> result its more messy than I thought.... and nearly impossible for me to check every record... feels better to have the data outside Wikidata...
  • The SBL dataset P3217 has as I said very very few errors P3217 (most errors I guess is me uploading it to Wikidata.... ;-)
    • The total dataset is 7756 records and with this query we just check birth dates and can see 16% has more dates with different precisions mostly because new uploads the last months....
  • Lesson learned
      1. ????
      2. Adding data to Wikidata and you wish to maintain the quality then you have a maintenance H#!?!?!?!ll ;-)
        1. I guess the upload rate will increase 10 times just the next year.....
  • Houston houston Wikidata has a problem.....
- Salgo60 (talk) 20:17, 16 December 2017 (UTC)
If you believe that data is that great, why don't you put it into the preferred rank? ChristianKl (✉) 21:10, 16 December 2017 (UTC)
  • In this case we don't want to discourage a person from adding the data. It's not the responsibilty of the person who adds the data to go to every example and decide what should have the preferred rank. ChristianKl (✉) 16:46, 16 December 2017 (UTC)
  • The phrase " having constraint checks for mass upload " suggest to me a poor understanding of our infrastracture. There's no single "mass upload" feature. We have many tools that can create mass uploads and for every tool you can discuss of how it should talk to the user about constraints. With a tool like QuickStatements I don't think it would be valuable to tell the user about non-mandatory constraints. ChristianKl (✉) 13:51, 18 December 2017 (UTC)
New clause? We already have that, it's called SAMPLE(). See Wikidata:WikiProject sum of all paintings/Creator missing collection authority control on how to use it. Multichill (talk) 09:22, 17 December 2017 (UTC)
How do we motivate that wdt just returns one value for other data types but for dates with different precisions in the database we get more...
SAMPLE is one way and then you need start grouping etc... --> queries get more complex. For me a good design is return the date with best precision for best rank and not more rows - Salgo60 (talk) 12:29, 17 December 2017 (UTC)
I would like to have the values returned with the most reliable source. This is not necessarily the value with the best precision. What is the adavantage of having precise but probably wrong values compared to having less precise but reliable values? --Pasleim (talk) 12:59, 17 December 2017 (UTC)
But ranking with the above logic is the responsibility of the person doing the query ==> you will often have entries with
  • same rank
  • different precision
today you get every entry with the same rank and all precisions back in theory you could get for Napoleon (Q517) birth the following returned
11 - day: 15 augusti 1769
10 - month: augusti 1769
9 - year: 1769
8 - decade: 1760
7 - century: 1700
6 - millennium: 1000
....
and then what do we do with different calendars ;-)
my suggestion is if everyone has the same rank pick best precision
- Salgo60 (talk) 13:14, 17 December 2017 (UTC)
wdt will always give multiple values when there are multiple values in the database for a given statement and there's no ranking information that gives information about which value happens to be best.
If cases where the user cares about data quality that's the correct behavior because it allows the user to decide for themselves which value they trust the most. In your case you are saying that data quality isn't the priority but you only want one value in every case because you are looking at too much data. That might be valid for your use-case but there's no reason for making it the default. Showing multiple values when there are multiple values on the same rank also encourages the user to actually make a judgement about which value is best and that's helpful for Wikidata.
In the case you linked above the most precise value was at the time you linked imported from the Russian Wikipedia. We consider that to be a source that's less reliable than Swedish Musical Heritage/Nationalencyklopedin/Musikverkets auktoritetsdatabas which existed as sources for the less precise claim. If a human would set a preferred rank that rank shouldn't be set to the Russian Wikipedia value. ChristianKl (✉) 13:32, 17 December 2017 (UTC)
The user case is: As an Wikidata user writing queries using fields with datetime I would like to have a simple way retrieving dates that is not fragile because of mass upload with data with odd precisions.
The key question I feel is
  1. Napoleon is born in the 1700 century
  2. Napoleon is born 15 aug 1769
Does those statements show 2 different values or one value presented with two different precisions. I vote for one value...
If you comoare those statements with this article then you also have another value 1768 for his birth
We will have the same "problem" with coordinates if people start uploading more precise or less precise coordinates...-- Salgo60 (talk) 21:13, 17 December 2017 (UTC)
It's interesting that you make up a fictional example to make your point given that you said you spend 30 hours dealing with real examples.
You ignore the issue of mass-uploads data with higher precision but lower quality.
When it comes to coordinates the whole thing becomes more complicated and it's even less clear that high precision should be the criteria to pick between different values. GeoNames data has high precision and that doesn't imply that data that might have higher quality should be hidden from the user just because it's lower precision. ChristianKl (✉) 13:47, 18 December 2017 (UTC)

Subitem Volume by periodical items

Consider Q45747062 now classified as a periodical. However this is a specific issue (6 december 1799) and many similar items will follow (9 issues, see Commons:Category:Journal de Bruxelles) Two issues are digitalized in fr:Wikisource. At the higher level Journal de Bruxelles (1790-1800))) is a separate item. To confuse matters there have several periodicals with the same name. The is about another periodical started in 1820. The two level distinction is made in the Wikisources so Wikidata needs to do similarly.

It seems to me that items about periodicals need to be ordened into 'periodicals' and 'volume/issue of periodicals' for the lower level. A better name can be found in English and other languages (In en:Wikisource 'Volume' is often used).Smiley.toerist (talk) 19:36, 17 December 2017 (UTC)

@Smiley.toerist: In the periodical publishing I'm familiar with, "issue" and "volume" are two different things, though both define a subset of the periodical. An issue is in principle one bound print object, published on a certain date, while a volume would be a collection of issues published over some period of time (often one volume per year). We have the properties issue (P433) and volume (P478) to indicate the issue number and volume number of a periodical in which a certain work appeared, but I don't believe we have anything to specifically show that an item is an issue or volume of a given periodical. I'm not sure particular issues in general are notable, but I guess some must be, so perhaps we do need a new property (or more than 1) for this? ArthurPSmith (talk) 14:45, 18 December 2017 (UTC)

Wikidata weekly summary #291

 
Here's your quick overview of what has been happening around Wikidata over the last week.
  • Development
    • Additional languages for monolingual values got approved by the language committee: ami (Amis), bnn (Bunun), fos (Siraya), ppu (Papora / Hoanya), pwn (Paiwan), pyu (Puyuma), ssf (Thao), trv (Seediq / Taroko), uun (Pazeh) (see gerrit:374052, phab:T144272)
    • More work on caching constraint results so we can enable the constraints checks for all logged-in users by default
    • Fixed a bug when edit links showed up in diff views (phabricator:T181807)
    • Final step towards being able to store statements on Forms of a Lexeme
    • Disabled RDF support for Lexemes as the mapping isn't defined yet (phabricator:T182660)
    • Adding some more smart tracking for Lua usage so we better know which properties from an item are used on a Wikipedia article even when the whole item is loaded (phabricator:T179923)

You can see all open tickets related to Wikidata here.

Wikipedia:Articles for deletion/List of Mayors of Wildwood, New Jersey

This might be of interest to people here at Wikidata: w:Wikipedia:Articles for deletion/List of Mayors of Wildwood, New Jersey, a dozen lists are up for deletion under this AFD. The discussion is about whether Wikipedia should retain lists of mayors of places in the United States based on a particular cutoff in population; or if they were appointed or elected. It is also an important issue of whether data should be preserved by merging and redirecting, or removed completely through deletion. --RAN (talk) 17:49, 18 December 2017 (UTC)

Precision of date

If I create a statement with date, for example publication date (P577) with a value of 2015, precision year, it shows 2015. But if I create a query, it shows for the same value Jan 1, 2015. Did I create the date wrong or the query service doesn't understand the precision? Or I should make a difference query? :) Samat (talk) 13:47, 17 December 2017 (UTC)

How about providing an example of the item you are talking about and the query you are making? ChristianKl (✉) 13:52, 17 December 2017 (UTC)
You are right. Example query. Samat (talk) 14:10, 17 December 2017 (UTC)
  • ?item wdt:P585 ?date yields 2015-01-01T00:00:00Z without any further information about precision
  • ?item p:P585/psv:P585 [ wikibase:timeValue ?time; wikibase:timePrecision ?precision; wikibase:timeTimezone ?timezone; wikibase:timeCalendarModel ?calendarModel ] . contains full details. The date bound to ?time still has the format 2015-01-01T00:00:00Z as before, but the ?precision variable with value 9 now tells you that this is to be interpreted with year-precision. Doc is at mw:Wikibase/Indexing/RDF Dump Format#Predicates for Time.

—MisterSynergy (talk) 14:07, 17 December 2017 (UTC)

Thank you! I think I will need some time now to understand what does this mean for the example query (where some data have year, some month and some day precision) :) Samat (talk) 14:15, 17 December 2017 (UTC)
Is this really how the system is supposed to work? Why doesn't the query service honor the precision? ChristianKl (✉) 12:08, 18 December 2017 (UTC)
Yes it is. The Query Service (intentionally) outputs raw data, which in particular means that the output of dates always has the same format "+2015-01-01T00:00:00Z"^^xsd:dateTime regardless of any precision in order to be processable by scripts. Since the date value itself does not contain any information about its precision, you have to ask for this information explicitly and format the date output properly by yourself with your script (the same applies to timezones and calendar models; for timezones I am not sure whether we can actually set values different from 0, but the feature is available in the data model). The Query Service can’t really do this due to the same reason. It only converts "+2015-01-01T00:00:00Z"^^xsd:dateTime to something more human-readable like Jan 1, 2015 in the web frontend, but at this moment it has no idea about the precision of this value. —MisterSynergy (talk) 12:45, 18 December 2017 (UTC)
That sounds to me like the web frontend is doing a poor job of parsing the value. I created a ticket https://phabricator.wikimedia.org/T183135 ChristianKl (✉) 13:00, 18 December 2017 (UTC)
As I said, output is intentionally raw and this is desired behavior. It is the Query Service user’s responsibility to format and/or interpret data correctly, and in fact all required information is available for each and every date. Users of the Query Service web frontend have basically three options:
  1. Don’t care about precision (use the wdt: prefix)
  2. Care about precision (use the p:/psv: prefixes, output both date and precision value, then interpret data notionally)
  3. Care about precision (use the p:/psv: prefixes and invent some ugly SPARQL hacks that output the date “properly formatted”)
I explicitly advise against #3 in most cases. —MisterSynergy (talk) 13:10, 18 December 2017 (UTC)
I think it's likely that someone is going to use "wdt:" and think that if the he's shown is "Jan 1, 2013" that this implies that it's actually the first Januar and the fact that he's shown the first Januar isn't just an artifact of the formatting. ChristianKl (✉) 21:01, 18 December 2017 (UTC)

A question about the Darts

Hi - in recognition of the current 2018 PDC World Darts Championship (Q30646688) I thought I'd take a look at how this is captured in Wikidata, and it's raised a few questions. So am I right in thinking the item PDC World Darts Championship (Q2634432) should be an instance of (P31) a recurring sporting event (Q18608583)? In addition I note that 2018 PDC World Darts Championship (Q30646688) is given an instance of (P31) PDC World Darts Championship (Q2634432). Is this correct? Would it also be sensible to give 2018 PDC World Darts Championship (Q30646688) an instance of (P31) tournament (Q500834)? Both recurring sporting event (Q18608583) and tournament (Q500834) are sub-classes of sports competition (Q13406554) so they would seem to be distinct items. That being said would it make sense to create a new item "Darts Tournament" which would be a sub-class of tournament (Q500834). This would allow queries that could specifically pull out darts tournaments, which couldn't quite be done at the moment. I'm not sure the way things are organised at the moment are quire right, so before making any changes it would be good to get input. Any advice welcome. JerryL2017 (talk) 19:15, 18 December 2017 (UTC)

British Museum

Hello. I have the feeling that we might need to change our 3 properties related to the British Museum (Q6373) – British Museum person or institution ID (P1711), British Museum thesaurus ID (P3632) and British Museum place ID (P3633) – because they have apparently changed their formatter URL (P1630) but also possibly the IDs that we store. Are they still working for you? Thierry Caro (talk) 03:11, 15 December 2017 (UTC)

What's up with organizations destorying their links lately? Sjoerd de Bruin (talk) 14:10, 15 December 2017 (UTC)
Please create new properties if they actually changed the identifiers.
--- Jura 20:52, 16 December 2017 (UTC)
OK. This seems to be working again, eventually. Thierry Caro (talk) 07:33, 19 December 2017 (UTC)

Brazil Weather Data

Greetings!

My name is Caio Silva, I'm a Information Systems student in Universidade Federal Fluminense, Rio de Janeiro, Brazil. I would like to upload some data about the weather in Brazil from a weather institute. They collect data everyday in every state of Brazil. I mailed them asking for permition to upload their data. IF, they let me use their data, I would like to upload them here, to be available for everyone using wikidata and wikipedia. Can you help me do it? I don't really know how to procede.

Thanks!

Sincerely, Caio Silva

@Caiosilva: Is there any plan to use that data ? Daily data is too big for WD which doesn't have the structure to store that kind of data. Snipre (talk) 08:24, 18 December 2017 (UTC)
This is what I said: WD is not able to store that information and data have to be uploaded in commons. Snipre (talk) 10:31, 19 December 2017 (UTC)

Wikidata fulltext search prototype

Hi!

Search Platform team would like to present a prototype test site of new and improved Wikidata fulltext search:

http://wikidata-wdsearch.wmflabs.org/wiki/Special:Search

Please try your favorite searches on it and report whether it looks good and which problems you notice. Important to note for this prototype:

What kind of problems we are looking for?

  • Ranking and retrieval problems, i.e. result X appears too low or too high in specific search, or does not appear at all (please tell us specific search query and expected result)
  • UI problems - i.e. the ranking is fine but highlighting or label or description is broken or look bad, or not highlighting the result that should be highlighted

Of course, if some search result worked spectacularly better for you, it would be nice to know too.

What should work?

Any search in Special:Search in main namespace and Property namespace should produce sensible result. Searches without advanced syntax should have better results than before, and search with advanced syntax (+, -, *, quotes, etc.) should work no worse than before.

Please note that this is a test wiki, so nothing else but search is expected to work, including clicking on other links, editing, browsing to other pages, etc. This is also a test site, so short disruptions might be possible when we update or change things or fix bugs reported by you :)

How to provide feedback?

 – The preceding unsigned comment was added by ‎Smalyshev (WMF) (talk • contribs) at 20:34, 18 December 2017 (UTC).

Comments

The time is the same as before - IIUC it's last edit time. I'm not sure how useful it is, but it didn't change from before. Smalyshev (WMF) (talk) 08:29, 19 December 2017 (UTC)
I searched for "P:result" http://wikidata-wdsearch.wmflabs.org/w/index.php?search=P%3Aresult&title=Special:Search&profile=default&fulltext=1&searchToken=epwo08wvhj1avsyhbjslg7f0g . It would be desireable to rank results (P2501) higher, but it's not a big deal.
I am not sure I understand you right - did you type in "P:result"? Why did you type "P:" in front of it? Smalyshev (WMF) (talk) 08:29, 19 December 2017 (UTC)
Strike that, I got it - it's for namespace searching. Smalyshev (WMF) (talk) 08:32, 19 December 2017 (UTC)
A while ago I typed "P:homo" to find sexually homologous with (P4545) which didn't work with the old search and also doesn't work with the new one. The same goes gor "P:death" finding number of deaths (P1120).
Searching "Heart Rhythm" with the search on the top finds cardiac rhythm (Q5038333) as second result. The old search gave it place #8 and the new one doesn't show it in the top 500 search results. This seems to me like a flaw that has to be fixed. When searching an item like cardiac rhythm (Q5038333), the user often doesn't know how the actual Wikidata item is called and might type the name of an alias. Hundreds of academic papers and/or musical songs shouldn't rank better. It's worth noting here that it's likely that the academic papers will be even better interlinked in the future with cites work (P2860).
Aliases should be ranked the virtually the same as labels. European Heart Rhythm Association/Heart Failure Association joint consensus document on arrhythmias in heart failure, endorsed by the Heart Rhythm Society and the Asia Pacific Heart Rhythm Society (Q26796261) however has three matches in the label, so we could not really rank it lower... Unless we rank all scientific articles lower. If you think this is necessary, please submit a phab task, I think it would be the best way. Smalyshev (WMF) (talk) 08:29, 19 December 2017 (UTC)
I created a ticket https://phabricator.wikimedia.org/T183243 , additionally it might make sense to rank exact matches with labels or aliases higher. ChristianKl (✉) 14:10, 19 December 2017 (UTC)
I would prefer "P:cites" to bring up cites work (P2860) as the first result. It might be valuable to make non-external-id properties rank higher than external-id properties. ChristianKl (✉) 22:29, 18 December 2017 (UTC)
I am not sure why cites work (P2860) ranks lower than CITES Species+ ID (P2040), maybe the latter has more usage? More usage usually produces higher rank. I looked and P2040 has 18261 usages, while P2860 has 1465587... So it should rank higher. Not sure what's up, I'll check it. Smalyshev (WMF) (talk) 08:29, 19 December 2017 (UTC)

Dams and reservoirs

I am very happy to see the new property reservoir created (P4661), but I'm not sure about has facility (P912) as its inverse. Currently dam (Q12323) is not <subclass of> facility (Q13226383) and I don't think of a dam as a facility in English. Thoughts? - PKM (talk) 20:11, 18 December 2017 (UTC)

I support removing that constraint, it wasn't raised during the property proposal process and if there really was an inverse we probably wouldn't have recommended to create the property. ArthurPSmith (talk) 16:22, 19 December 2017 (UTC)

George Wallace (Q313776)

Hi! Can someone with more knowledge of George Wallace (Q313776) take a look at his position held (P39)? Is it correct as is or more to do? //Mippzon (talk) 09:36, 19 December 2017 (UTC)

Seems to have been fixed by @Oravrattas now. Thanks! //Mippzon (talk) 12:41, 19 December 2017 (UTC)
It would be good to also add the end time (P582) to his member of the Alabama House of Representatives (Q18648747) statement but I don't know what that should be. (electoral district (P768) would be a useful qualifier there too.) I've also added a candidacy in election (P3602) for his presidential run in 1968, and the Democratic Party presidential primaries in 1964, 1972, and 1976. --Oravrattas (talk) 13:19, 19 December 2017 (UTC)
Awesome! Wallace has a pretty extensive "profile" on Wikiadata now, good job! The template on Swedish Wiki which uses Wikidata-data looks pretty nice with your additions: Link //Mippzon (talk) 17:11, 19 December 2017 (UTC)

Events that start and end in same year

Should they be listed as "starttime=1917 and endtime=1917" or as "pointintime=1917", I see a mix of both. --RAN (talk) 14:06, 19 December 2017 (UTC)

I prefer "starttime=1917 and endtime=1917". I see "pointintime=1917" as saying that we know that the event happened inside the given year but that doens't preclude it from also happening the year before or after. ChristianKl (✉) 14:09, 19 December 2017 (UTC)

New dashboard on the Wikidata Concept Monitor: Geo dashboard

 
Map of the 1000 libraries of Wikidata that are most used in the Wikimedia projects

Hello all,

You may have heard about the Wikidata Concept Monitor (WDCM) website, a tool full of statistics about Wikidata's content and usage. It was released for the 5th birthday by Goran, data scientist in the Wikidata team.

A new board is now available on the platform: the Geo Dashboard. It creates a map of the 1000 most used items in several semantic categories: museums, libraries, art galleries, universities, botanical gardens, etc. The usage of the items being the count of the number of pages in the Wikimedia projects where the Wikidata item is used.

Feel free to have a look at this board and the others, check the documentation and ask any questions if you have some. You can also suggest other categories that you would like to see represented on the Geo Dashboard.

Cheers, Lea Lacroix (WMDE) (talk) 15:28, 19 December 2017 (UTC)

Meistersingerhalle

... is the same as this: No label defined (Q41427480), please merge. --Gerda Arendt (talk) 13:45, 22 December 2017 (UTC)

  Merged if you meant Meistersingerhalle (Q1734355) and Meistersingerhalle (Q41427480). Matěj Suchánek (talk) 13:57, 22 December 2017 (UTC)
This section was archived on a request by: Matěj Suchánek (talk) 09:58, 25 December 2017 (UTC)

automatic tools for languages

Hi, I'm sorry if this is not right place to add this question. is there any automatic tool, script or gadget to make sitelink automaticlly between many languages instead add it manually.

For Example: I put list of articles in french and list of the English articles or item(Q) (in order), and the tool link between them --Ibrahim.ID (talk) 12:25, 24 December 2017 (UTC)

If I understand you correctly, quickstatements is a tool which can help. It takes tab-separated lists of item, property, value and makes the edits. You need a WiDaR login to use Quickstatements. Help:QuickStatements is your friend, from which:
  • To add a sitelink to a specific page on a site to an item, use "Sxxx" instead of the property, with "xxx" as the site (e.g. enwiki, commonswiki).
    Example: Q340122 TAB Szhwiki TAB "塞浦路斯·諾爾維特"
    Meaning: add sitelink to Chinese Wikipedia (塞浦路斯·諾爾維特) to Cyprian Norwid (Q340122)
hth --Tagishsimon (talk) 17:43, 24 December 2017 (UTC)
yes, this is exactly what I want, Thanks @Tagishsimon: --Ibrahim.ID (talk) 19:30, 24 December 2017 (UTC)
This section was archived on a request by: Matěj Suchánek (talk) 09:58, 25 December 2017 (UTC)

Undo merge Cholla empire

Hoi, a dynasty and an empire are quite distinct. I cannot unmerge it.. https://www.wikidata.org/w/index.php?title=Q151148&action=edit&undoafter=590211633&undo=598540062 Thanks, GerardM (talk) 15:10, 24 December 2017 (UTC)

What exactly do you want to undo?--Ymblanter (talk) 22:11, 24 December 2017 (UTC)
@GerardM:   Done --Liuxinyu970226 (talk) 02:29, 25 December 2017 (UTC)
Thank you :) GerardM (talk) 05:28, 25 December 2017 (UTC)
This section was archived on a request by: Matěj Suchánek (talk) 09:58, 25 December 2017 (UTC)

Help with “Criterion used”

I’d like confirmation that I’m using criterion used (P1013) correctly. I want to say that if spinning jenny (Q49015) and water frame (Q1751504) are <subclass of> textile working tools/equipment, as Getty AAT says, then so are spinning frame (Q3274449) and spinning mule (Q1413723) by analogy. Are the qualifier and reference on the subclass statement at spinning mule (Q1413723) the right way to do this? - PKM (talk) 00:52, 17 December 2017 (UTC)

I don't feel that it's clear what information you want to express in that item. ChristianKl (✉) 19:39, 17 December 2017 (UTC)
Yeah, I was afraid of that. Will consider whether I am overthinking this. Perhaps the subclass statement is so obvious that it doesn’t need a convoluted justification. - PKM (talk) 19:35, 19 December 2017 (UTC)

Try the prototype to improve the term box

 
Screenshot of the changes for the Term box

Hello all,

The so-called “term box” is the part of the item page where the labels, descriptions and aliases are displayed in different languages. We are currently working on improving the interface for editing this section, especially regarding adding new languages.

We just published a prototype, and we would really like your input so we make sure that our changes will fulfill your needs. You can have a look at the prototype, and answer a few questions on the input page. It will take you about 10 minutes, and will really help us to understand how you expect to edit the labels in multiple languages. We will collect the answers until January 11th.

If you have any question or suggestion, feel free to write on the talk page. Thank you very much, --Hanna Petruschat (WMDE) (talk) 13:54, 20 December 2017 (UTC)

Data model for importing UNESCO's 'number out of school children'

I'm getting ready to start a data import from UNESCO institute of statistics, using the property number of out-of-school children (P2573) to add the data to country items in Wikidata. There are lots of facets to the data which would be amazing to have represented in Wikidata, but there is a serious risk of over cluttering country items with lots of stats so I wanted to get some input from the community.

We have the data broken down as follows:

Age:

  • Primary age
  • Lower Secondary age
  • Upper Secondary age
  • All ages combined

Gender:

  • Male
  • Female
  • both sexes

Time:

  • New data is released annually going back to 1999.

To me, a good starting point would be to add the 'all out of school children' data (i.e. total number of out of school children, of both sexes combined). It should also include all the historical values (using point in time (P585) qualifier). This would allow us to query the data to show the changes over time.

I really feel like we should have also have the data for separated 'male' and 'female' children recorded as well, as this will highlight serious gender inequality issues. This would mean including 2 additional P2573 statements with qualifiers like 'applies to part' = 'female' / 'male' (or similar). If we also include historical values, each country could have up to 36 statements + another 18 statements for 'all out of school children', and up to 3 new statements added per year.

Does anyone feel this is going to be too much to throw into country items? Should we just stick with 'all children', despite missing out on important data about gender inequality? Many thanks, NavinoEvans (talk) 15:59, 18 December 2017 (UTC)

  • I don't think worry about too much data on country items should be a driving concern. If there's really too much data we will have to find ways to not load all data at once but collapse the view of subsets of the data of an item. ChristianKl (✉) 17:08, 18 December 2017 (UTC)
Thanks for this, that sounds like something we will need at some point in the future anyway. I'll wait a bit for any other objections but otherwise will go ahead with the import. Cheers, NavinoEvans (talk) 09:26, 19 December 2017 (UTC)
  • I thought the general idea was to add such data as tabular data to Commons. Items that have too many statements can become hard to load. As country items are used in many applications, I'm not sure if this is a suitable place to add the data.
    --- Jura 10:31, 19 December 2017 (UTC)
I see the question of whether data should be on commons is about whether it's valuable to have the data within the reach of SPARQL. When it comes to daily temperatures data for every city in existance having that data in SPARQL is too expensive even when some people could do useful queries. I don't think storing yearly data about 'number out of school children' for countries would have the same problem. ChristianKl (✉) 14:18, 19 December 2017 (UTC)
The above lists a lot of datapoints. It's just a matter of capacity. I think eventually the query service was said to include structured data on Commons.
--- Jura 14:29, 19 December 2017 (UTC)
The query service would have had a relatively hard time searching terabytes worth of climate data on commons and I personally can't remember a plan that SPARQL was supposed to include WikiCommons. ChristianKl (✉) 14:35, 19 December 2017 (UTC)
phab:T181319 - Support external tabular datasets in WDQS. --Yair rand (talk) 01:47, 20 December 2017 (UTC)
Very interesting indeed, I had assumed it would never be possible to access Commons tabular data with the Wikidata Query Service but at least it seems like it will be tried out at some point. I still think we should add the yearly data broken down by gender at this stage because it's not clear if/when the power to query tabular data will arrive. Many things will need adjusting at that time anyway as some properties will need changing or removing completely I presume. NavinoEvans (talk) 15:20, 20 December 2017 (UTC)
That would be just 3 per year? np.
--- Jura 17:16, 20 December 2017 (UTC)
Yes that's right, many thanks! NavinoEvans (talk) 12:30, 21 December 2017 (UTC)

Merge Florianópolis Q IDs

Please merge Florianópolis (Q32124221) to Florianópolis (Q132997), they refer to the same entity. | Jaider Msg 16:54, 19 December 2017 (UTC)

ceb.wikipedia begs to differ, with w:ceb:Florianópolis versus w:ceb:Florianópolis (munisipyo). --Tagishsimon (talk) 17:27, 19 December 2017 (UTC)
This instance is part of a bigger issue. Cebuano Wikipedia (Q837615) and Swedish Wikipedia (Q169514) have separate articles for the municipality (Q15284) and the locality (Q3257686) of the same name. But the articles for the settlement are often incorrectly linked to the Wikidata item of the municipality. A comprehensive approach to fix this problem is needed. --MB-one (talk) 12:34, 20 December 2017 (UTC)
Should note that the issue originates from the fact that GeoNames (Q830106) has this duplication, and ceb is nothing but a bot-import of that (and other) database. While technically, the political unit isn't the same as the settlement - e.g. there might be political units which contain several settlements, the whole ceb/geonames import is a big mess. Ahoerstemeier (talk) 16:11, 20 December 2017 (UTC)
Or it just exposed the mess Wikipedia created. Sjoerd de Bruin (talk) 19:23, 20 December 2017 (UTC)

How to display results of a query as map on Wikidata/Wikipedia?

I've got a query (this one for example) whose results are nicely displayed as a map in WQS. Is there a way to embed this map in Wikipedia or Wikidata? Thanks. — Ayack (talk) 13:12, 20 December 2017 (UTC)

Maybe like fi:Wikiprojekti:Kartat/Esimerkkejä#Haku_Wikidatasta,_alueet_OpenStreetMapistä_2? Stryn (talk) 15:31, 20 December 2017 (UTC)
Can't find a way to display geopoints (your example is only about geoshapes). — Ayack (talk) 10:53, 21 December 2017 (UTC)
I think that only works on Wikivoyage.
--- Jura 11:11, 21 December 2017 (UTC)

Political prisoner

Good morning! What property can I use for political prisoner (Q217105)? significant event (P793) with qualifiers or occupation (P106) wich I find interpret as occupation of a person; see also "field of work" (Property:P101), "position held" (Property:P39) Breg Pmt (talk) 06:10, 10 December 2017 (UTC)

Political prisoner is a very POV title. In my country someone I consider a terrorist and mass-murderer is often considered a Political prisoner by my enemies. I recommend using prisoner (Q1862087) under significant event (P793) to steer clear of the POV issue. DGtal (talk) 06:34, 10 December 2017 (UTC)
We have place of detention (P2632) to mark where someone was imprisoned. subject has role (P2868) could be used to express that the person is a political prisoner but that should only be done with sources that clearly say so. ChristianKl (✉) 11:52, 11 December 2017 (UTC)
With place of detention (P2632) together with location (P276) and (from - to dates) as qualifiers for subject has role (P2868) as "main" property ? Breg Pmt (talk) 23:08, 11 December 2017 (UTC)
I don't see why you would need location (P276) but otherwise, I agree. ChristianKl (✉) 22:15, 13 December 2017 (UTC)
@ChristianKl: Like for Otto Heinrich Frank (Q7336) how can I indicate that he was prisoned in Auschwitz (Q7341) located in Germany, away from where he lived?. Pmt (talk) 06:05, 16 December 2017 (UTC)
You can use place of detention (P2632) Auschwitz (Q7341) and additionally you can use residence (P551) to state where he lived previously. Of course adding qualifiers helps with getting the data clearer. ChristianKl (✉) 12:55, 16 December 2017 (UTC)
  • Political prisoner is a well defined term. It does not matter what you call someone, it matters how the definition fits. So look up what a neutral organisation like Amnesty uses as a definition. Thanks, GerardM (talk) 12:02, 22 December 2017 (UTC)

What is the difference between motorized bicycle (Q1377570) and moped (Q201783)?

Aren't both of them simply bike with motor? C933103 (talk) 08:45, 17 December 2017 (UTC)

Not in Germany, there's a huge difference: One is a bicycle with an attached motor (and it has to have pedals, so it can be used without the motor), the other is a small motorbike with 50cm³. motorized bicycle (Q1377570) don't require a driving licence (or any street legality (Q228760) or de:Hauptuntersuchung, moped (Q201783) require all of this. Sänger (talk) 09:39, 17 December 2017 (UTC)
Same in UK, motorized bicycle (Q1377570) and moped (Q201783) are different things. JerryL2017 (talk) 09:53, 17 December 2017 (UTC)
So if I understand correctly, if you put a pair of pedal onto moped (Q201783), then it would become motorized bicycle (Q1377570), and then you would no longer need driving license?
And what would be the difference between them in places where they fall into same regulatory category?
C933103 (talk) 11:36, 17 December 2017 (UTC)
So I'm not sure about regulations in different countries, I guess they will vary. Conceptually though I would say the difference is that a motorized bicycle (Q1377570) can be ridden using the pedals even if the motor isn't running. The motor assists the rider but isn't required. Whereas a moped (Q201783) will always require the motor, even with pedals attached it wouldn't really be possible to propel the vehicle any distance without the motor driving the wheels. JerryL2017 (talk) 12:55, 17 December 2017 (UTC)
Then, how about pedelec (Q2066172)? It seems to me a bike with optional motor power too.C933103 (talk) 13:12, 18 December 2017 (UTC)
@C933103: Unfortunatelly there's actually no possibility for any items that are mentioned in this section to be merged (should we really run mergism game here?), even Q1377570 has not only Q2066172, but also e.g. natural gas (Q40858)-powered bicycles, or photovoltaics (Q192127)-powered... --Liuxinyu970226 (talk) 10:04, 22 December 2017 (UTC)
@Liuxinyu970226: I'm not trying to merge these items, just want to assess what's the difference between these items in order to ensure their entry in different language versions are properly linked. C933103 (talk) 12:02, 22 December 2017 (UTC)

novalue and format constraint violation

Hi, if a property has no value and we want to specify the non existence by setting the value of a statement with novalue, we get a violation constraint if only a string format was defined. How can we modify the format constarint to accept both format constraint for a value and novalue as value ? Thanks Snipre (talk) 21:12, 20 December 2017 (UTC)

Can you give an example where that's an issue? ChristianKl (✉) 01:24, 21 December 2017 (UTC)
@ChristianKl:Two examples:
Snipre (talk) 08:56, 21 December 2017 (UTC)
The reports are generated by User:KrBot. Only the bot operator can change this. —MisterSynergy (talk) 09:06, 21 December 2017 (UTC)
@Ivan A. Krestinin: Ivan, can you exclude novalue statements from your format check analysis ? Snipre (talk) 10:15, 21 December 2017 (UTC)
Given that we likely soon migrate the constraint away from being bot based to being done natively, solving the problem at that time might be more efficient. ChristianKl ❪✉❫ 15:48, 21 December 2017 (UTC)
Well – we would like to offer a replacement for Wikidata:Database reports/Constraint violations in the WikibaseQualityConstraints extension (phabricator:T180582), but it probably won’t happen within the next half year. At least until then the bot-based system will still be necessary, and we should try to make the two systems consistent with each other. --Lucas Werkmeister (WMDE) (talk) 18:52, 21 December 2017 (UTC)

Efficent way to check if item is a instance/subclass of something

In a Javascript-Userscript I want (for example) to check if the current item instance of (P31) software or a subclass of (P279) of a software. My current solution is: 

const qid = mw.config.get().wbEntityId,
      endpointUrl = 'https://query.wikidata.org/sparql',
      sparqlQuery = 'Ask { wd:' + qid + ' wdt:P31/wdt:P279* wd:Q7397.}',
      fullUrl = endpointUrl + '?query=' + encodeURIComponent( sparqlQuery ),
      headers = { 'Accept': 'application/sparql-results+json' };
fetch( fullUrl, { headers } ).then( body => body.json() ).then( json => {
    console.log(json.boolean)
} );

It takes about half a second. Is there a more performant, easier or prettier way? -- MichaelSchoenitzer (talk) 02:13, 22 December 2017 (UTC)

The query is somewhat faster with an explicit hint to the optimizer: ASK { wd:Q83 wdt:P31/wdt:P279* wd:Q7397. hint:Prior hint:gearing "forwards". }. The optimizer should figure this out on its own, but currently it doesn’t (phabricator:T168973).
(Also, unrelated side note: using mw.config.get( 'wbEntityId' ) seems to be somewhat faster than mw.config.get().wbEntityId too – the latter seems to load all of wbEntity on demand, which even triggers a deprecation warning.) --TweetsFactsAndQueries (talk) 12:37, 22 December 2017 (UTC)

+300K humans in just one week

Hello. I noticed that Wikidata counts 300,000 additional humans in just one week. This tool counts the number of items in Wikidata with instance of (P31) = human (Q5) (and nothing else), without checking when they were created. We can see that this set of 300,000 humans seems to come without any property like sex or gender (P21), country of citizenship (P27), date of birth (P569), and occupation (P106). Does someone know what happened? — Envlh (talk) 08:35, 22 December 2017 (UTC)

Tell us what could be improved with the entity suggester

Hello all,

We are currently looking into improving the search tool behind the search field (top right of each page) and the entity suggester (for example when you edit a statement). We are aware that some of the suggestions that appear and their ranking in the list is not always what you expect, and we would like to know more about this.

This page collects examples of current behavior of the suggester. I already added the first suggestions from ChristianKl. Feel free to add some more, so we know what are the most important troubles you encounter.

Thanks, Lea Lacroix (WMDE) (talk) 10:09, 22 December 2017 (UTC)

target for merge of non-human person (Q18170708)?

This "non-human person" at ca/esWPs would seem to have a match somewhere in our plethora of persons. At the moment it is sitting without items and needs a home or a merge IMO.  — billinghurst sDrewth 14:57, 17 December 2017 (UTC)

non-human (Q13023682)? --Fractaler (talk) 15:48, 17 December 2017 (UTC)
Also, non-human electoral candidate (Q2740853), for example, is "non-human person"? --Fractaler (talk) 15:55, 17 December 2017 (UTC)
What do they mean with "person" in ca/es? Do they count corporations as persons? ChristianKl (✉) 16:19, 17 December 2017 (UTC)
Legally, they are "juridical persons". You can see juridical person (Q155076) Climent Sostres (talk) 22:34, 22 December 2017 (UTC)
Note: Users who edited Q18170708: @Climent Sostres, Andreasmperu: ^^ --Liuxinyu970226 (talk) 04:18, 18 December 2017 (UTC)

Excessive football team rankings?

A bot has been adding hundreds of rankings per item to football teams such as San Marino national football team (Q185350) and Chinese Taipei national football team (Q459119). This seems excessive to me. I could understand listing the yearly rank, but adding hundreds seems overboard. The bot's user page does not link to a relevant bot permission request, so I'm assuming this has not been discussed already. At what point are claims too granular (or numerous) to go in Wikidata? Kaldari (talk) 19:18, 21 December 2017 (UTC)

@Pasleim: ^. Kaldari (talk) 19:20, 21 December 2017 (UTC)
My bot only helped to correct values which were before added by User:Xaris333 with QuickStatements. For doing QuickStatements edits, our policies don't require previous discussions. --Pasleim (talk) 20:15, 21 December 2017 (UTC)
There should be a property "tabular ranking" that takes the bulk out of those items. I have suggested that @Xaris333: propose it, but he has continually declined. Mahir256 (talk) 20:22, 21 December 2017 (UTC)
Note that the community of Czech Wikipedia is interested in having this property in the infoboxes. I'm in favor of keeping this until tabular data can be loaded from wikis. Matěj Suchánek (talk) 20:30, 21 December 2017 (UTC)
  1. First, I have not declined anything. I have just said "I don't know anything about Tabular Data. I can't understand what is it."
  2. Note, that the also Greek wiki is interested in having this property in the infoboxes. So, I aggree with Matěj Suchánek.
  3. Note that FIFA publish ranking every month. There is no a yearly rank, a rank in the end of the year.
  4. My addition is not something new. Users were adding the same data with a P2656 (P2656) Wikidata:Requests for deletions/Archive/2017/Properties/1#Property:P2656. I have just added the whole data, from 1992 to 2017, with sources.

Xaris333 (talk) 15:13, 22 December 2017 (UTC)

I don't think that Wikidata should include the whole ranking history. Is it even allowed by FIFA? Why not only show the latest and the best ranking? Stryn (talk) 16:40, 22 December 2017 (UTC)
Why not? Ranking is a ranking. Now we have December Ranking. Should we delete it in January to add the new one??? And why only the latest and the best? Why not the worst? And if a national team has the 2nd place once in all their history, and 20 times the 3rd place, why only to have the one time they have the 2nd place? Why not to have the latest and the best population of a place? Why to have it all? I have more questions and examples... It's the first time I am reading an opinion that said that Wikidata should select some data of a data list... (And what the problem with FIFA? Many data we have in Wikidata are from organizations). Xaris333 (talk) 17:05, 22 December 2017 (UTC)
  • On the problem with FIFA: we are only allowed to import data here that is properly licensed, i.e. compatible to Wikidata’s CC0 license. Not sure what FIFA’s license for the imported data is, but typically sports organizations are not at all open for permissive licensing of their data. At least FIFA sits in Switzerland where the level of protection for databases is not very high, so this complicated and restrictive EU database directive does not apply here.
  • On the completeness: given an import was allowed, I do not see a problem in holding full data—somewhere in the Wikimedia cosmos. Right now Wikidata is very likely the best (and only) place to do so, but in future tabular data at Commons might be a better solution for bulky data such as this one. Right now, it is practically impossible to use the data from there, so this is not yet an option.
—MisterSynergy (talk) 17:15, 22 December 2017 (UTC)
Where can I read for properly licensed in Wikidata? And we can't use data if we add the source with it? Xaris333 (talk) 17:23, 22 December 2017 (UTC)
meta:Wikilegal/Database Rights and the Wikipedia links therein are probably the best we have right now. If other editors have more input, I’d be interested as well. —MisterSynergy (talk) 17:26, 22 December 2017 (UTC)
Hm, reading this is seems that many database are not allowed to used in Wikidata. Like population data of a country. Xaris333 (talk) 17:41, 22 December 2017 (UTC)
  • As a general rule of thumb I would advocate a standard where yearly data is okay and higher resulting goes to WikiCommons but when there a need for the infobox to have higher resolution, why not? ChristianKl ❪✉❫ 20:31, 22 December 2017 (UTC)

Exporting labels

Is there any easy way to export all entered labels by language?

For example, if I am creating a new "fictional human" item, I'd like to use the labels from fictional human (Q15632617) as default descriptions via QuickStatements. When (if?) we have automatic descriptions generated from statements, I assume it will do this, but I'd like something to use in the meantime.

I suppose a gadget that creates default descriptions from instance of (P31) and subclass of (P279) (or both if both exist) would be even easier. Any suggestions? - PKM (talk) 20:03, 22 December 2017 (UTC)

Issue with search function?

I have noticed an issue with some recent items which return "no match found" in the "Search Wikidata" box - using both their label and their Q-ID - although they clearly exist. This also makes it impossible to set some property's value to them. Examples are Pierre Fouchet (Q45825730), Pochi (Q45825741), district mayor of Tiergarten (Q45825742) or Dörre Wieslein (Q45825750). The affected items are not continuous in ID (e.g. Erick Guerrero (Q45825749) works fine). The issue persists both when switching browsers and when logging out.--Pfadintegral (talk) 11:14, 17 December 2017 (UTC)

I can replicate the issue. ChristianKl (✉) 12:33, 17 December 2017 (UTC)
This seems to be the issue from phab:T183101. I've updated all affected items I could find manually, hopefully this does not repeat. Looking into how prevent it from happening again. Smalyshev (WMF) (talk) 01:26, 22 December 2017 (UTC)
Thank you - however, of the ones I mentioned, Pochi (Q45825741) and district mayor of Tiergarten (Q45825742) still seem to be affected to me. --Pfadintegral (talk) 13:12, 23 December 2017 (UTC)

Dyes and pigments

Sometime in the new year, I want to do some work on colourants, dyes, and pigments. I'll probably rely on the hierarchy in Getty AAT unless someone knows of a better or more expanded controlled vocabulary.

I don't think we've done much on mapping relationships such as Indigofera tinctoria (Q165213) <has product> indigo (Q422662) <contains> (or ?<source of>) indigotindisulfonate sodium (Q410120) <colour produced> blue (Q1088).

  1. Does anyone have experience mapping these relationships?
  2. Is there support for creating a new property <colour produced> and new qualifier <mordant used>? This would not be the same as color (P462) since color (P462) should be the colour of the dyestuff itself, which may produce a range of colours with different mordants. (We may also want a way to say that "woad" ?<overdyed with or ?<in combination with> "madder" produces "purple".) - PKM (talk) 20:38, 22 December 2017 (UTC)
There is Indigofera tinctoria (Q165213) <this taxon is source of> indigo (Q422662) and the inverse indigo (Q422662) <natural product of taxon> Indigofera tinctoria (Q165213). And this already is in place. - Brya (talk) 12:14, 23 December 2017 (UTC)
Creating the property would be a discussion best had on property proposals. A few pointers: Why "produces" as a word? If that's the best word, why not a more generalized property? ChristianKl ❪✉❫ 12:56, 23 December 2017 (UTC)

Contributions

Hoi, when I look a my contributions the setting of "Hide probably good edits" is always on. This makes contributions of no value to me. I use it to know where I left of in a project when I am at another computer. Please fix this nuisance. Thanks, GerardM (talk) 07:33, 23 December 2017 (UTC)

In Special:Preferences#mw-prefsection-rc, you have to untick the option "Show only likely problem edits (and hide probably good edits)". --Pasleim (talk) 08:58, 23 December 2017 (UTC)
in the contributions it does not stick. It is where I expect to make this change. Thanks, GerardM (talk) 10:26, 23 December 2017 (UTC)
On the right side there's a save icon next to the deletion icon. ChristianKl ❪✉❫ 13:00, 23 December 2017 (UTC)
No there is not. We are talking about where it matters, in the contributions page. GerardM (talk) 14:39, 23 December 2017 (UTC)
"Hide probably good edits" should always be automatically unticked, except if you have chosen "Show only likely problem edits (and hide probably good edits)" in your preferences. Stryn (talk) 15:21, 23 December 2017 (UTC)

What would the ideal data import process look like?

 

We are running an RFC on creating a better and easier data import process, we are mapping the current including tools and documentation, what is available and what is still needed. Please take a look here.

Many thanks

John Cummings (talk) 14:53, 23 December 2017 (UTC)

Merge castle park (Q2607557) and castle park (Q2244647)?

Can castle park (Q2607557) and castle park (Q2244647) be merged? To me they are the same, but I would appreciate someone else's eyes on it! //Mippzon (talk) 17:07, 28 December 2017 (UTC)

  Done - They were indeed about the same concept. Mbch331 (talk) 17:39, 28 December 2017 (UTC)
Thanks! /Mippzon (talk) 18:23, 28 December 2017 (UTC)
This section was archived on a request by: Matěj Suchánek (talk) 09:15, 29 December 2017 (UTC)

What should be instance of (P31) of Alexander Hamilton and slavery (Q27958361)

What should be instance of (P31) of Alexander Hamilton and slavery (Q27958361) item? --Jarekt (talk) 19:52, 21 December 2017 (UTC)

political position (Q20045456). Sjoerd de Bruin (talk) 19:58, 21 December 2017 (UTC)
Thanks--Jarekt (talk) 16:49, 22 December 2017 (UTC)
@Jarekt: "Wikipedia article". It is most certainly not a "political position". Andy Mabbett (Pigsonthewing); Talk to Andy; Andy's edits 17:48, 22 December 2017 (UTC)
Andy, I am confused, are you suggesting instance of (P31) of Alexander Hamilton and slavery (Q27958361) item should be Wikimedia article page (Q15138389) (or something similar)? If so, than large fraction of items on Wikidata would also fall in the same category and I do not think it would be proper use of instance of (P31) property. I usually try to make sure every item I look at has instance of (P31) or subclass of (P279) and Q27958361 did not; however I did not know what to label it as. political position (Q20045456) suggested by User:Sjoerddebruin seems like a good category. Why do you object to it? --Jarekt (talk) 18:04, 22 December 2017 (UTC)
Why not create a new item "personal political position" as a subclass of political position (Q20045456) ? That would allow people to distinguish the abstract and general from personal and particular. Jheald (talk) 18:11, 22 December 2017 (UTC)
"Abolitionism" is a political position. "Alexander Hamilton and slavery" is not. Andy Mabbett (Pigsonthewing); Talk to Andy; Andy's edits 20:02, 22 December 2017 (UTC)
I based myself on Special:WhatLinksHere/Q20045456. Sjoerd de Bruin (talk) 15:52, 23 December 2017 (UTC)
Q2915249 is more relevant. Andy Mabbett (Pigsonthewing); Talk to Andy; Andy's edits 12:28, 24 December 2017 (UTC)
It should certainly not be "Wikipedia article". That's not the topic; the linked articles do not start with the text "The Wikipedia article 'Alexander Hamilton and slavery' was created in October 2016 by...". The article has a topic, and it is Alexander Hamilton's relationship with slavery, and not some Wikipedia article. --Yair rand (not logged in) 22:31, 24 December 2017 (UTC)
So shall we remove all instances of "Wikimedia list article" and "Wikipedia disambiguation page", then? Andy Mabbett (Pigsonthewing); Talk to Andy; Andy's edits 23:23, 24 December 2017 (UTC)
No. There exist some pages in the main namespace of Wikipedia that are not encyclopedia articles. These pages frequently have titles that reference what the page is, rather than what it describes. Examples include lists, indexes, outlines, disambiguation pages, and glossaries (eg. w:Glossary of glass art terms, which is not about a glossary, but is a glossary itself). These pages' associated items, like those for categories/templates/etc, exist primarily for interwikis, but they also make use of certain properties for tracking purposes, for the sake of convenience, even though it is not how the property should normally work. P31 is used in these items, acting as though the item is about the linked pages themselves in such situations. (In my opinion, we should really use a separate dedicated property for that, but since we don't yet have one, P31 works for the time being.) --Yair rand (not logged in) 04:47, 25 December 2017 (UTC)

Javascript: best way to add statements?

What's the best way to add a statement in a userscript? So far I found only the quite annoying manual way: 

// add data
api = new mw.Api();
api.get( { action: 'query', meta: 'tokens'}).then(
    function(aw) {
        token = aw.query.tokens.csrftoken;
        api.post( { 
            action: 'wbcreateclaim',
            entity: qid,
            property: pid,
            snaktype: 'value',
            value: '"string"',
            token: token
            }).then(
            function(aw) {
                console.log(aw.claim);
                api.post( { 
                    action: 'wbsetclaim',
                    claim: JSON.stringify(aw.claim),
                    token: token
                }).then(
                   function(aw){
                     //...
                   });       
            });
    } );

I guessed that jQuery.wikibase.snakview is the way to go, but I havn't jet understood how it works and the documentation is spare. Is there a minimal working example? In general I miss a resource for getting started with the JS-Api. I'd wish for something like Wikidata:Creating a bot for Userscripts. Is there such a thing already or could we create it? -- MichaelSchoenitzer (talk) 15:06, 22 December 2017 (UTC)

In general I miss a resource for getting started with the JS-Api. That’s because we don’t (officially) have a JS API. jQuery.wikibase.snakview is Wikibase-internal, and (along with all other Wikibase JavaScript code) not covered by our Stable Interface Policy. (Neither is the HTML DOM structure, for that matter.) You can try to use it, but your user script might break at any time when Wikibase changes. We’d like to offer stable interfaces to gadgets and user scripts at some point, but we don’t have anything like that yet. Sorry for the bleak reply :/ --Lucas Werkmeister (WMDE) (talk) 16:45, 22 December 2017 (UTC)
Thanks Lucas for the info! I now did it via the api and still hope to find a better solution one day. -- MichaelSchoenitzer (talk) 00:44, 24 December 2017 (UTC)

Rice variety

I need a help. Bangladesh Rice Research Institute (BRRI) [Q4855372], an agricultural research institute in Bangladesh work on development modern variety rice. BRRI already developed so many rice varieties. I want to know, Can I added the variety list at Wikidata? If yes, an instance of the area, which one is best to use? --Hasive • talk • 16:09, 28 December 2017 (UTC)

instance of rice (Q5090) ? Snipre (talk) 21:37, 28 December 2017 (UTC)
@Hasive: You might model that on varieties of apple, like Mabbott's Pearmain (Q41779076). Andy Mabbett (Pigsonthewing); Talk to Andy; Andy's edits 19:52, 29 December 2017 (UTC)
Thanks Andy Mabbett and Snipre. --Hasive • talk • 12:14, 30 December 2017 (UTC)
This section was archived on a request by: Matěj Suchánek (talk) 21:12, 31 December 2017 (UTC)

Merge recreation area (Q2063507) and recreation area (Q338112) or not?

To me those to items: recreation area (Q2063507) and recreation area (Q338112) seems very similar. Are they the same and can be merged, or are they different in some sense? //Mippzon (talk) 09:48, 29 December 2017 (UTC)

@Fralambert: as this user marked both together with different from (P1889). --Liuxinyu970226 (talk) 13:12, 29 December 2017 (UTC)
The first one is something that has been built, it has subclasses like amusement park (Q194195). The second one is something that is protected, it is a subclass of protected area (Q473972). just like bird sanctuary (Q2714144) is. Not to be merged if you ask me, as a specific object can be an instance of just one (or both) of these objects. --Larske (talk) 16:30, 29 December 2017 (UTC)
I support the merge. The US recreation areas I am familiar with contain camp sites, built trails, visitor centers, and picnic areas. I agree that they are protected, but they are also built up for human recreation. - PKM (talk) 20:15, 29 December 2017 (UTC)
There is no problem to have these US recreation areas to be an instance of (P31) both classes if you see fit. But you can't have amusement park (Q194195), or anything else that is not in all cases a protected area, as a subclass of (P279) the merged object if the merged object is still to be a subclass of (P279) protected area (Q473972). I don't think that the amusement park (Q194195) Disneyland (Q181185) or Fun-World (Q5508621) are protected area (Q473972), i.e. "protected areas". So then you have to find a new "superclass" to amusement park (Q194195). --Larske (talk) 03:49, 30 December 2017 (UTC)
Thanks for the clarification! With this reasoning, I don't think the items should be merged either. //Mippzon (talk) 09:37, 30 December 2017 (UTC)
Good point.- PKM (talk) 19:30, 30 December 2017 (UTC)
This section was archived on a request by: Matěj Suchánek (talk) 21:12, 31 December 2017 (UTC)

Byte count

Quick question out of curiosity. Why does a revision like this one (just changing a letter of a label) increases the byte count of the page by 146 according to the revision history? I've seen this happening sometimes.--Ssola (talk) 17:07, 29 December 2017 (UTC)

Because the way how we save the content changes from time to time, bytes don't mean a lot on Wikidata. Sjoerd de Bruin (talk) 17:11, 29 December 2017 (UTC)
Ok, thank you!--Ssola (talk) 18:00, 29 December 2017 (UTC)
This section was archived on a request by: Matěj Suchánek (talk) 21:12, 31 December 2017 (UTC)

Merge proposal

Hey all, I believe that items consonantal alphabet (Q27948966) (consonantal alphabet) and abjad (Q185087) (abjad) are to be merged because they represent the exact same concept, but there are two articles in Esperanto (and only in Esperanto). Is the only way to proceed to contact WP:eo and propose the articles to be merged there? If so, is someone here active on the esperanto Wikipedia ?--Alexandre Hocquet (talk) 03:09, 25 December 2017 (UTC)

If eoWP considers them different concepts then they are by default different items here. If you can convince eoWP to merge the items as the same concept at that point they can be merged. Otherwise they would be Wikimedia permanent duplicate item (Q21286738)  — billinghurst sDrewth 06:34, 25 December 2017 (UTC)
@Andreasmperu:, one of our sysops, in fact aganist this merging. However, as neither their Wikidata user page (see ping link), nor Meta-Wiki and nor eswiki (kindly, in their native speaking language) indicate that that sysop has (even -1 or -0) Esperanto knowledge, maybe asking native speakers @ArnoLagrange, Iketsi, Kani, Schneelocke: would be better in order to make correct judges. --Liuxinyu970226 (talk) 15:06, 25 December 2017 (UTC)
Given that we also have different items for alphabet and writing system I don't see how the two are the same thing. A writing system is something that's more broad than just an alphabet. A writing system for example includes a direction in which it's written and read. ChristianKl ❪✉❫ 15:57, 25 December 2017 (UTC)
consonantal alphabet (Q27948966) is alphabet (Q837518). "string of consonantal alphabet" is string (word) formed from objects/letters of this alphabet (subset of the set Q42558923). abjad (Q185087) (group of all "well-formed formula (Q655328) consonantal alphabet") is alphabet (Q837518)? --Fractaler (talk) 08:46, 26 December 2017 (UTC)

Festivals that does not fully depend on calendar

For Christmas (Q19809), we can use day in year for periodic occurrence (P837) to denote it take place on December 25 (Q2745) (or other dates for other places).

For Easter (Q21196), we have used day in year for periodic occurrence (P837) with earliest date (P1319) and latest date (P1326) to denote the range of dates the festival can possibly occur.

However, Is it possible to specify the exact calculation rule for the property? As in, is it possible to make wikidata include the data that Easter (Q21196) is a festival that occurs during the First (Q351401) Sunday (Q132) from the First (Q351401) Ecclesiastical full moon (Q5332270) since March 21 (Q2433) (for Christian churches)?

Likewise, is it possible to denote that Setsubun (Q746798) occurs 1 (Q199) day (Q573) before Lichun (Q1364685), Cold Food Festival (Q715157) occurs 105 (Q716840) day (Q573) after Dongzhi (Q1679621) (for Korea), or that Coming of Age Day (Q483226) occurs on 2 (Q200)nd Monday (Q105) in January (Q108)?

C933103 (talk) 21:09, 25 December 2017 (UTC)

I don't think we have currently an established way of doing this. If you have a proposal about what kind of properties you would want to model this, I encourage you to create a property proposal. ChristianKl ❪✉❫ 13:06, 26 December 2017 (UTC)

Wikidata class structure tool?

Hi! Are there any available tools that can display the classes of items in a tree structure? Like which item is subclass of (P279) to another etc. //Mippzon (talk) 10:39, 26 December 2017 (UTC)

I think you are looking for this Tool https://tools.wmflabs.org/wikidata-todo/tree.html? --McSearch (talk) 10:45, 26 December 2017 (UTC)
Well, that's a great tool! But I'm not interested in what items has the instance of, I just want to know the relationships between classes. For instance I want to visualize that amusement park (Q194195) is subclass of (P279) to leisure park (Q15982170). Maybe I'm not using the correct terminology though :) //Mippzon (talk) 10:53, 26 December 2017 (UTC)
I like this external tool for such tasks. --MisterSynergy (talk) 11:10, 26 December 2017 (UTC)
There is also https://lucaswerkmeister.github.io/wikidata-ontology-explorer/ --Lydia Pintscher (WMDE) (talk) 11:11, 26 December 2017 (UTC)
Thanks! This is more of what I was looking for! Thanks all! //Mippzon (talk) 11:34, 26 December 2017 (UTC)
Internal tools: 1) Template:Item documentation --Fractaler (talk) 13:20, 26 December 2017 (UTC)

Wikidata weekly summary #292

 
Here's your quick overview of what has been happening around Wikidata over the last week.
  • Development
    • Due to the Christmas and New Year's Eve holidays, no deployment will happen before January 3rd.
    • Monolingual language codes for Old French (fro) and Middle French (frm) added (phabricator:T181823)
    • Finishing touches on storing statements on Forms
    • No longer checking constraints on the property example statements (phabricator:T183267)
    • More work on caching constraint check results so we can roll out constraint checking for all logged in users
@Lydia Pintscher (WMDE): One of the significant issues that were named in the slides about Wikidata quality was broken links. Wikipedia solves this problem through the collaboration with the Internet Archive. What would need to happen for us having a bot that does the same task? Would it make sense to approach the internet archive and ask for collaboration? ChristianKl ❪✉❫ 12:42, 26 December 2017 (UTC)
Good point. I was talking to them a while ago already about this and they said they're working on making their bot work for Wikidata. I'll be in San Francisco in January and will see if I can get another meeting and status update on it. --Lydia Pintscher (WMDE) (talk) 12:44, 26 December 2017 (UTC)

Scientific names

Why is validly published name (Q23729754) marked as an instance of nomenclature code (Q2673092)? And the latter, as an instance of biological nomenclature (Q522190)? Andy Mabbett (Pigsonthewing); Talk to Andy; Andy's edits 22:16, 17 December 2017 (UTC)

It isn't. --Succu (talk) 23:14, 17 December 2017 (UTC)
What isn't? Andy Mabbett (Pigsonthewing); Talk to Andy; Andy's edits 00:17, 18 December 2017 (UTC)
@Pigsonthewing: The former of your concerns has not been true since September. @Brya:, as the last person to introduce the statements of concern. Mahir256 (talk) 02:55, 18 December 2017 (UTC)
Thank you, Mahir256. It seems I was inadvertently looking at an old ID for for the former item. The later relationship, though, is still a concern. Andy Mabbett (Pigsonthewing); Talk to Andy; Andy's edits 15:30, 18 December 2017 (UTC)
Mind to voice your „concern” a little bit more? --Succu (talk) 22:51, 19 December 2017 (UTC)
As Succu points out, it is not, although really it should be. The relationship validly published name (Q23729754) as an instance of nomenclature code (Q2673092) seems fine: it is informative. The present relationship "subclass of" "name" is quite uninformative. - Brya (talk) 03:21, 18 December 2017 (UTC)
name (Q82799) is a homonym. validly published name (Q23729754) is the scientific name (scientific term (Q999234)?) of the object. --Fractaler (talk) 08:10, 18 December 2017 (UTC)
No. validly published name (Q23729754) is applicable to name a taxon (Q16521). --Succu (talk) 22:46, 19 December 2017 (UTC)
So, validly published name (Q23729754) is a scientific term (component of scientific terminology (Q13419784))? Now we have: validly published name (Q23729754) is second-order class (Q24017414) (second-order class (Q24017414) is third-order class (Q24017465) is fourth-order class (Q24027474) is fixed-order class (Q23959932) is variable-order class (Q23958852) is class (Q23960977) is variable-order class (Q23958852) is class (Q23960977) is (and here we have a cycle)). --Fractaler (talk) 10:42, 20 December 2017 (UTC)
No, it's not a "scientific term" like e.g. anode (Q181232), it's a name governed by the respective Code. --Succu (talk) 22:56, 21 December 2017 (UTC)
Ok, this "a name governed by the respective Code" is scientific or non-scientific/unscientific? --Fractaler (talk) 10:04, 22 December 2017 (UTC)
Your "alternatives" make not a lot of sense to me. A nomenclature code (Q2673092) is a set of rules you have to follow if you want to publish a new name. If you fulfill the requirements you have published a validly published name (Q23729754) in accordance with the governing Code. There is no science involved, it's simply following the law. Applying such a name to a circumscription of a taxon should follow scientific principles. --Succu (talk) 21:12, 22 December 2017 (UTC)
Ie, as a result, we have now names that either follow scientific principles, or do not follow. Correctly? --Fractaler (talk) 17:58, 23 December 2017 (UTC)
No. You are mixing nomenclature and taxonomy. --Succu (talk) 18:56, 23 December 2017 (UTC)
Ok, we'll figure out who and what is mixing now. For example, Homo sapiens sapiens (Q3238275) is taxon (Q16521). taxon (Q16521) is second-order class (Q24017414). Homo sapiens sapiens (Q3238275) is second-order class (Q24017414)? --Fractaler (talk) 07:39, 26 December 2017 (UTC)
What's your point? --Succu (talk) 07:51, 26 December 2017 (UTC)
There are no editorial points of view. For example, WP considers that: "H. s. sapiens is the only extant human subspecies" Fractaler (talk) 09:00, 26 December 2017 (UTC)
To repeat the question of Succu: What's your point? Sänger (talk) 09:30, 26 December 2017 (UTC)
And for what do you point of view of Wikidata editor (Q28859215)? WD is not built on the principle "This Wikidata editor (Q28859215) thinks so, so the superset of this set X is X superset". If you are interested in what links in my opinion are better for WD, then this, of course, is scientific. --Fractaler (talk) 11:19, 26 December 2017 (UTC)
Yes, but what exactly is your point? And how is this related to this topic? --Succu (talk) 21:37, 26 December 2017 (UTC)
I just want to understand what you're talking about. Above you said:"Applying such a name to a circumscription of a taxon should follow scientific principles". I took a concrete example - Homo sapiens sapiens (Q3238275). Now Homo sapiens sapiens (Q3238275) is taxon (Q16521). taxon (Q16521) is second-order class (Q24017414)...etc...is mathematical object (Q246672). So, "Applying such a name to a circumscription of a mathematical object should follow scientific principles"? And now validly published name (Q23729754) is name (Q82799) (word or term used for identification). name (Q82799) can be 1) scientific 2) unscientific/non-scientific. So, I just want to know what validly published name (Q23729754) is 1) or 2). --Fractaler (talk) 08:13, 27 December 2017 (UTC)

We seem to have been sidetracked; and (to return to my original point) we are still claiming - without any given source - that nomenclature Codes (description: "lawbooks of nomenclature, in biology") are an instance of biological nomenclature (description: "collective term for various naming systems"). Andy Mabbett (Pigsonthewing); Talk to Andy; Andy's edits 20:25, 26 December 2017 (UTC)

Above I asked you to to voice your „concern” a little bit more. You didn't. Maybe you have an alternative proposal, how the relationship should be expressed? --Succu (talk) 21:32, 26 December 2017 (UTC)

Properties

Hello. I want to know if there are some properties to use for en:Template:Infobox person: (Or other way to have these data to Wikidata)

  • date of baptism (P1636) is about the date. Do we have place of baptism?
  • date of disappearance (P746) is about the date. Do we have place of disappearance? Do we also have "disappeared status"?
  • Do we have body_discovered property? About the place.
  • place of burial (P119) is about the resting place. Do we have property about burial place?
  • influenced by (P737) is about influenced by. Do we have a property influence? (Aristotle influence Alexander the Great.)
  • convicted of (P1399) and penalty (P1596). Do we have a property about criminal status? If the person is still in jail?
  • award received (P166) is about award received. In also know as it said "honours". So, we are using awards to add both awards and honors?
  • Do we have a property about significant monuments erected, buildings named, etc., in honour of the subject?

Xaris333 (talk) 21:07, 25 December 2017 (UTC)

@ChristianKl:

  • "How do you see the burial place to be different as the resting place?" The templates means that burial place is the place where the body is burial. Resting place it the location of columbarium, ash-scattering, etc. Is there a way to have that data?
  • Do we have a property from the place where a person was raised and matured, if different from birthplace and residence?
  • How we add the TV show a presenter presented?

Xaris333 (talk) 19:03, 26 December 2017 (UTC)

In most cases you have either a buried body or a location of a columbarium. If there's a need to be more precise maybe you could use object has role (P3831).
I don't think that we have a different one but residence can be qualified with start/stop time.
I'm not sure about the TV shows, maybe someone else is more clear on that front. ChristianKl ❪✉❫ 20:37, 26 December 2017 (UTC)
for the presenter of a TV show, you can add presenter (P371) on the TV show item. --Pasleim (talk) 13:15, 27 December 2017 (UTC)

Thank you! Xaris333 (talk) 15:55, 27 December 2017 (UTC)

Punctuation and aliases

Consider, for example, John Patton O'Neill (Q21388086), which is labelled in English as "John Patton O'Neill" (with a straight, ASCII apostrophe). Should that have the alias "John Patton O’Neill" (with a single curly quote), or would that be redundant? Andy Mabbett (Pigsonthewing); Talk to Andy; Andy's edits 20:50, 26 December 2017 (UTC)

Yes and no. Bitwise they would be different, but they have the same meaning. From a very pedantic point of view the simplified (straight) typewriter apostrophe would be redundant to the true (curly) typographic apostrophe, but only in this direction---which is typically not found here (in English labels we use straight typewriter characters as enwiki does). --MisterSynergy (talk) 22:30, 26 December 2017 (UTC)
Aliases are only useful for searching. If the new search engine can handle such queries, it will be redundant. Matěj Suchánek (talk) 09:58, 27 December 2017 (UTC)

User:Egor Shustoff

I've just overview contributions of this user and see that he's merged many article items with disambig items. It's too much to me to revert them. Someone will help? --Infovarius (talk) 13:22, 27 December 2017 (UTC)

Should amusement park (Q194195) use inception (P571) or date of official opening (P1619) to show when the park opened the first time?

I'm having a hard time figuring out if amusement park (Q194195) should use inception (P571) or date of official opening (P1619) for its first opening? Using date of official opening (P1619) for amusement park (Q194195) gives a constraint violation, so is it more accurate to use inception (P571)? //Mippzon (talk) 09:52, 25 December 2017 (UTC)

I think date of official opening (P1619) is better but at the moment the superclasses of amusement park (Q194195) aren't well develeoped. The solution that takes the least work would be to have amusement park (Q194195) subclass fixed construction (Q811430) but it might be well possible to find a more clever way to think about the related subclasses if someone wants to spend more effort. ChristianKl ❪✉❫ 12:12, 25 December 2017 (UTC)
I see! How can we develop the superclass of amusement park (Q194195)? Is there a process for it, or is it just to define some instances of amusement park (Q194195) and have them as examples? //Mippzon (talk) 12:28, 25 December 2017 (UTC)
We don't have a formal process. You can just add other classes and if nobody objects they stay. If you want to put in effort it can be useful to search for prior art in other controlled vocabularies to see how they classify it and create references. ChristianKl ❪✉❫ 12:42, 25 December 2017 (UTC)
You might want to use the Getty AAT hierarchy for “amusement park” to build the superclasses. I do this a lot with material culture items, and cite the AAT item as a reference for the <subclass of> statements. - PKM (talk) 19:58, 25 December 2017 (UTC)
Thanks, great link! But I have a hard time understanding how to translate this structure to Wikidata. Can you give an example to help me out? //Mippzon (talk) 10:32, 26 December 2017 (UTC)

┌────────────────────────────────────────────────────────────────────────────────────────────────────┘ Sure! We already have building complex (Q1497364) (AAT 300000202). We also have recreation area (Q338112), which seems to be about recreation areas in the US, and recreation area (Q2063507) which is about recreation areas in other places. Personally, I don't see a distinction between these items. I would suggest removing the "different than" statements, merging recreation area (Q338112) and recreation area (Q2063507), assigning AAT 300000510 to the merged entity, and making it a subclass of building complex (Q1497364) using AAT as the reference. Then I'd make amusement park (Q194195) a subclass of the new merged entity.

For an example of a different hierarchy based on AAT in Wikidata, see this tree. - PKM (talk) 19:42, 26 December 2017 (UTC)

That sounds like a good idea! Since I'm not used to work on this "high" level in Wikidata, is this something that I can do myself or do I need consensus to apply those changes? Where can in that case get that consensus? //Mippzon (talk) 09:50, 27 December 2017 (UTC)
Merges are tricky to reverse, so if you’re unsure about a merge you can always ask here or at the relevant Wikidata project, if there is one. Other than that, filling out class hierarchies (especially where you have a reputable source to use as a reference) is something you can do on your own. Just poke around and make sure there isn’t one already using some other wording. There’s a lot of gaps between high level concepts and lower level physical things in our hierarchies, so this sort of work is really helpful. - PKM (talk) 05:24, 28 December 2017 (UTC)

Population synthesis

Hello. Let say a town has 100 French and 80 German. So, the population is 180. We can have population (P1082) --> 180. How we can have population synthesis (100 French and 80 German). It is not only about nationality. Some census counted by religion etc. Xaris333 (talk) 22:11, 26 December 2017 (UTC)

Do you mean, has part(s) (P527) French quantity (P1114): 100, German: 80? --Fractaler (talk) 09:38, 27 December 2017 (UTC)
This is not a good solution since we are connected the data to a population property. Xaris333 (talk) 12:39, 27 December 2017 (UTC)
Perhaps we could use a similar scheme as with qualifiers male population (P1540)/female population (P1539) (and perhaps make those more general). Matěj Suchánek (talk) 09:54, 27 December 2017 (UTC)
And what should be the criteria to segregate this people? Some racist junk like in the american census? Some stuff like ethnic nationalism (Q637251)? Or the passports? I don't see any value in such artificial and arbitrary segregation of humans. Sänger (talk) 11:47, 27 December 2017 (UTC)
Nothing racist. If a census count persons these ways, Wikidata can have those informations. For some countries are historical informations that connected with countries historical facts. Xaris333 (talk) 12:37, 27 December 2017 (UTC)
My ancestors did such disgusting racist stuff as well, and then put some of this arbitrary categories in the gas chambers, the South-African apartheid criminals did so as well. The categories of the american census are just useless junk, based on imho criminal racist distinctions. ethnic nationalism (Q637251) and everything related is just a crime. Sänger (talk) 12:47, 27 December 2017 (UTC)
@Matěj Suchánek: We can't have qualifiers for all. Male/female are easy, are only two categories. Xaris333 (talk) 13:15, 27 December 2017 (UTC)
One property is enough. We can use units. Matěj Suchánek (talk) 13:19, 27 December 2017 (UTC)
Example? population (P1082) --> 180. Qualifiers male population (P1540)--> 70, female population (P1539) --> 100. Then? How we are going to show how many Frenchs and how many Germans? Xaris333 (talk) 14:06, 27 December 2017 (UTC)
Qualfiers: "population by nationality": 123 Germans (Q42884), 456 French (Q121842) ... Matěj Suchánek (talk) 14:13, 27 December 2017 (UTC)
@Matěj Suchánek: Oh, you mean to have a new property. Fine with me. What about "population by religion"? And why not to have just "population by gender" and not two properties male population (P1540)/female population (P1539)? Xaris333 (talk) 15:32, 27 December 2017 (UTC)
"population by religion" would work the same way. Good point on the gender as well.
I also suggest to change the scope of those properties: from less general "population" to "number"/"portion". This would make the properties useful in other contexts than population, like number of deaths (P1120). Matěj Suchánek (talk) 16:26, 27 December 2017 (UTC)
@Matěj Suchánek: So you suggest to have "number by gender", "number by nationality", "number by religion". But, how this will work. Example, population (P1082) --> 180. Qualifiers PNUMBER BY NATIONALITY (Pnumber by nationality) --> 100 French, PNUMBER BY NATIONALITY (Pnumber by nationality) --> 80 German. How French and German are going to add? By unit? Xaris333 (talk) 16:53, 27 December 2017 (UTC)
I think units would then be the only possiblity. In the end, it depends on the community whether this will be allowed. Matěj Suchánek (talk) 17:20, 27 December 2017 (UTC)
That is a problem. I can't propose a new property if it is not possible to be created. Xaris333 (talk) 17:45, 27 December 2017 (UTC)
The proposal can make this clear. Matěj Suchánek (talk) 18:52, 27 December 2017 (UTC)
Done. Wikidata:Property proposal/Number by nationality, religion and gender. Xaris333 (talk) 20:15, 27 December 2017 (UTC)
If you report all categories of the census by year for hundreds of years that will be a lot of data. I think it would make to sense to store this information as an additional row in tabular population (P4179) ChristianKl ❪✉❫ 12:46, 27 December 2017 (UTC)
It seems that soon the Wikidata will have a label "Wiki-structured data", and Commons -> "Wiki-unstructured data" --Fractaler (talk) 12:56, 27 December 2017 (UTC)
@Fractaler: tabular data on Commons is also structured data. The difference is that on Commons the data representation is 2d whereas on Wikidata it's 1d. --Pasleim (talk) 13:01, 27 December 2017 (UTC)
hierarchy (Q188619) is 1D? What is structured data While no one gave an exact definition, while only "information with a formal data model". And Commons has only >0D-data. --Fractaler (talk) 13:20, 27 December 2017 (UTC)

Google Code-in mentor request

I have set up a Google Code-in task for Wikidata content, in a fairly isolated area of Wikidata but relevant to Google Code-in students, but before we can publish it we will need a few more mentors to help me. https://phabricator.wikimedia.org/T182766

The task is adding facts about Google Code-In (Q5583820) organisations/projects. A list of them can be seen at http://gci-leaders.netlify.com/

Signup as a mentor at https://www.mediawiki.org/wiki/Google_Code-in/2017 ; more info at https://www.mediawiki.org/wiki/Google_Code-in/Mentors . An Org admin will then invite you to Google Code-in system.

Then add yourself as a mentor on the Phabricator task or on the GCI task (there are only five Wikidata tasks, so it is easy to find in the system :-( )

Part of the background to this task is that many GCI orgs try to get students to write Wikipedia articles as a task, often English Wikipedia articles, and it usually ends in tears. Their mentors dont know Wikipedia, so cant actually help the students. See facebook: WikipediaWeekly group post for my analysis of those types of tasks.

As we know, Wikipedia is hard, so rather than have students try to write articles, I believe they will have more success with more granular tasks such as Wikidata facts.

I've trialed this task in another GCI org, reasonably successfully so far. see coala association e.V. (Q45337328) and coala (Q25215462).

There are lots of very easy facts that can be added, but there are also lots of complexities. Every Open Source org / project has its own strangeness, especially in the relationship between them. e.g. on which item does the github username 'wikimedia' belong? ;-)

fwiw, I put out a similar request at facebook: WikidataCommunity, without any success so far. John Vandenberg (talk) 05:02, 28 December 2017 (UTC)

Use/meaning of items created by

Special:Contributions/Amadany. Nothing is linked to any project, and many items have no claims, or just a startdate/enddate combi. For me it all seems pretty clue-less. Edoderoo (talk) 06:05, 28 December 2017 (UTC)

And the reason you haven't posted on their talk page (currently a red link) is..? Andy Mabbett (Pigsonthewing); Talk to Andy; Andy's edits 10:31, 28 December 2017 (UTC)
I posted at Amadany's talk page, and they are now adding sub-classes to the new items they are creating. A reminder to all colleagues to be welcoming to and supportive of new users. Andy Mabbett (Pigsonthewing); Talk to Andy; Andy's edits 12:32, 4 January 2018 (UTC)
This section was archived on a request by: Andy Mabbett (Pigsonthewing); Talk to Andy; Andy's edits 12:32, 4 January 2018 (UTC)

Aliases of Homo Sapiens

One year after I last raised the matter, Homo sapiens (Q15978631) still lacks the (English) aliases "human", "man", "mankind", "woman", "womankind" and "people". Andy Mabbett (Pigsonthewing); Talk to Andy; Andy's edits 00:02, 21 December 2017 (UTC)

I see the talk page of the item as a better venue for having the discussion. As far as the substance of the argument goes I don't think it's a good idea to add those aliases as it will make it more complicated for new people who type "human" into the search bar to find the correct item given that the item we want them to use is in most cases human (Q5). ChristianKl (✉) 00:45, 21 December 2017 (UTC)
I agree with Christian. Sjoerd de Bruin (talk) 11:05, 21 December 2017 (UTC)
I can see lots of good reasons for including such valid aliases, at Help:Aliases. I see nothing there about deliberately omitting valid data as a tool for managing hypothetical mis-applicaton of items in Wikidata (the monitoring and repair of which in any case we have perfectly adequate tools). Also, the top of this page says that it is a "Place used to discuss any and all aspects of Wikidata: [including] individual data items". 22:11, 21 December 2017 (UTC)Andy Mabbett (Pigsonthewing); Talk to Andy; Andy's edits
Please reread and try to understand the comments given to you to your original post in 2016. --Succu (talk) 22:38, 21 December 2017 (UTC)
Succu, your snide comment, insinuating lack of understanding, where there is only justifiable disagreement, is inappropriate. But then, I credit you with the intelligence to have known that already. Andy Mabbett (Pigsonthewing); Talk to Andy; Andy's edits 23:21, 21 December 2017 (UTC)
If there are „lots of good reasons“ to do so, why aren't you give us some. --Succu (talk) 21:29, 22 December 2017 (UTC)
It's clear from your recent comments here that you have found both the previous discussion and Help:Aliases, and have read at least some of each of them. Andy Mabbett (Pigsonthewing); Talk to Andy; Andy's edits 13:33, 23 December 2017 (UTC)
Mind to offer your reasons why you think this aliases are helpful in this special case? --Succu (talk) 23:23, 23 December 2017 (UTC)
If you see good reasons and want to convince us, why don't you share your good reasons? As far as Help:Aliases goes, it doesn't explicitely forbid the labels you propose but the labels don't fit the inclusion criteria for things that should be included either. ChristianKl ❪✉❫ 19:58, 22 December 2017 (UTC)
Maybe Help:Aliases dating back to 2012 needs some overhaul? The "Scientific (binomial nomenclature) names" example Panthera atrox (Q242358) is not a good one. --Succu (talk) 22:43, 22 December 2017 (UTC)
Those inclusion criteria state, in part: "All of the other common names that an entry might go by". Andy Mabbett (Pigsonthewing); Talk to Andy; Andy's edits 13:37, 23 December 2017 (UTC)
It doesn't do that in the section of things that should always be included, with suggests that the guideline is not written to have always all possible common names included. ChristianKl ❪✉❫ 15:50, 23 December 2017 (UTC)
That section is quite clearly not intended to be exhaustive; but the text I quoted is directly under the heading "Inclusion criteria". Andy Mabbett (Pigsonthewing); Talk to Andy; Andy's edits 22:58, 23 December 2017 (UTC)
The text says that some things can be generally included and has a list that other things should be included. What's important to be included is under the should section. For those things that aren't in the should section it makes sense to see on a case by case basis whether it's valuable to include them, and the policy refers currently refers to the Project Chat for further discussion (I think the talk page would be the better venue). Here on the project chat you seem to be either unwilling or unable to provide arguments for why the aliases should be added in this case. ChristianKl ❪✉❫ 13:52, 24 December 2017 (UTC)
Given that you're replying to me doing so, your comment seems strange. Andy Mabbett (Pigsonthewing); Talk to Andy; Andy's edits 14:24, 24 December 2017 (UTC)
Intuitively, I'd say "human" should be an alias for "Homo sapiens": it is commonly used as such in science. The word "human" is interesting and has more than one meaning: there is the "a human" referring to an instance, and "human" as in "she is human". The latter is the content why I think at least "human" should be an alias for "Homo sapiens". I am less sure about man/woman mankind/womankind, which are likely very controversial at this moment: the meaning of these terms are in flux, and I mark them as controversial (for myself) and stay away from it at this moment. So,   Support for "human". --Egon Willighagen (talk) 14:44, 29 December 2017 (UTC)
@Egon Willighagen: The scientific statement "she is human" wouldn't be modeled as "she is homo sapiens" in Wikidata. Setting the label that way indicates that the item can be used that way and that's a wrong impression to give people who aren't familiar with our items. Why do you think misleading most naive users is worth adding the alias? ChristianKl ❪✉❫ 21:17, 29 December 2017 (UTC)
Who wants to set the label that way? In making such a claim, the misleading appears to be yours. and yours alone. Andy Mabbett (Pigsonthewing); Talk to Andy; Andy's edits 23:21, 29 December 2017 (UTC)
@ChristianKl:, it seems it was not clear that with "she is human" is argued for how the word human is used in common language. From your reply, it seems it suggested "she is human" was meant as statement in Wikidata. I would have used 'Qxxxx P31 "human"' in that case and am not at all arguing to change that. My point is that a common word like "human" has two meanings. I really do not think it is up to Wikidata to decide that one meaning is wrong. It's use for typing people is surely confusing as hell (e.g. with Homo (Q171283) that does have 'human' as alias, and all those music/bands/etc called 'human'); but I doubt it got much clearer now: I have to reread what I'm doing each time and I am aware of the issue. But Wikidata should not "guide" users by forbidding to use words as words are used; I think there are other mechanisms to do so, in fact, I believe a few of them are actually in place, that warn users about using the right "human" item. Misleading is wrong, naive users (I dislike that term) and experience users, alike. Since we're taking semantics to the next level now, humans are a bit special here... The same example applies with cat: "Kitty is a cat" and "Kitty is cat". The latter you would not say, and the first has a different meaning from "she is a human". There is a bit of linguistic constructs involved here, but it boils down to the instance and the class: person (Q215627) being the instance, Homo sapiens (Q15978631) being a class. Any specific human is an instance 'person'. This information should be what is guiding the "naive" human, and not forbidding the common use of words. BTW, 'human' seems a perfectly fine alias for Homo (Q171283) and has been used there for almost five years, without discussion. --Egon Willighagen (talk) 06:30, 30 December 2017 (UTC)
The alias for "homo" was never "human" but "Human" with obviously violated our norms as it's not a proper name, but it didn't get much attention. The fact that the obviously wrong label wasn't fixed isn't much argument in it's favor. It's easy for a new user to understand that an album or a song isn't the value they are seeking for when they read the description of the items. It's not as easy with "homo sapiens". ChristianKl ❪✉❫ 08:46, 30 December 2017 (UTC)

instance of (P31) of David in Islam (Q1257863)

Another interesting instance of (P31) question: Articles David in Islam (Q1257863) and Solomon in Islam (Q1768167) are for aspects of those 2 people that differ in Islamic tradition from the accounts in Judo-Christian tradition. Currently both instance of (P31) are human (Q5), but aspect (Q25351772) or something similar might be better. --Jarekt (talk) 17:55, 27 December 2017 (UTC)

I guess it should have facet of (P1269) and probably use the main item for most other properties. Similar issue with social media use by Barack Obama (Q4858119) --Jarekt (talk) 20:13, 27 December 2017 (UTC)
Barack Obama (Q76) is a person. What is person (Q215627)? person (Q215627) different from the "biography of a person" (biography (Q36279)) in Wikidata? From religious character (Q18563354)? It consists of real and virtual components (aspects, certain aspects of a person's life (Q20127274)). social media use by Barack Obama (Q4858119) is "virtual component of person"? Fractaler (talk) 10:26, 28 December 2017 (UTC)

Can "Basic membership properties" be updated with approach for products?

What is the current best practice for using instance of (P31) and subclass of (P279) for products or similar kinds of entities that are not physically unique in the world? I've read previous discussions debating this, however I think it would be very helpful if Help:Basic_membership_properties could be updated with some guidance about how products are handled. Products are collections, there are multiple "instances" produced when they are created. Fuzheado suggested the current thinking is that they are defined using subclass of (P279). Is that the general understanding? Pauljmackay (talk) 16:25, 29 December 2017 (UTC)

I think a good example (mostly) right now is aircraft (Q11436) - aircraft models or other groupings are subclass of (P279) aircraft (Q11436) (or a subclass of that), but specific models are also instance of (P31) aircraft model (Q15056995). A specific individual vehicle like Enola Gay (Q204424) is an instance of a model (Boeing B-29 Superfortress (Q184870) in this case) which is a subclass of heavy bomber (Q3534114) which is a subclass of (up the tree several levels) aircraft (Q11436). ArthurPSmith (talk) 17:01, 29 December 2017 (UTC)
@ArthurPSmith: could that be considered a solid guideline then for the community? Does anyone disagree with this general pattern? Pauljmackay (talk) 09:55, 30 December 2017 (UTC)
Small mass-produced design items are often considered instances. See Iphone7 and Windows7--Ssola (talk) 12:27, 30 December 2017 (UTC)
@Ssola: Can you expand on this? What is the reasoning? iPhone 7 (Q26831164) seems to be an example matching the aircraft example above, it is a subclass of iPhone and an instance of smartphone model (Q19723451). Windows 7 (Q11215) seems a bit more messy, its got a logical subclass but the "instance of" values overlap with the subclass and seem unnecessary IMHO. Pauljmackay (talk) 13:28, 30 December 2017 (UTC)
Sorry, I picked up bad examples. See Rubik's Cube (Q47043) instance of (P31) mechanical puzzle (Q924588). I don't advocate this use, I'm just pointing out this exists, and I think this should be corrected somehow.--Ssola (talk) 13:50, 30 December 2017 (UTC)

VIAF down?

Is VIAF down for other users too? I can't get through by any method I've tried, but don't know whether it's a general issue at their end, or merely at mine. --EncycloPetey (talk) 09:15, 30 December 2017 (UTC)

Same here. According to this page, they are currently performing upgrades of their infrastructure. "Maintenance for all data centers started at 2017-12-30 01:52 W. Europe Time planned until 2017-12-30 14:00 W. Europe". So it should be back online in a couple of hours. --Kam Solusar (talk) 10:10, 30 December 2017 (UTC)
In future you can check using http://downforeveryoneorjustme.com/ (or similar sites). Andy Mabbett (Pigsonthewing); Talk to Andy; Andy's edits 10:57, 30 December 2017 (UTC)

Born at sea and Died at sea

I asked once before how best to cluster all the people that were born-at-sea. I just created (birth at sea (Q46998262)) but I think it should appear as a qualifier for birth-location which is usually Atlantic Ocean or Pacific Ocean or Indian Ocean. Any ideas? --RAN (talk) 22:28, 28 December 2017 (UTC)

Why not just use wdt:P19 / wdt:P361* wd:Q715269 ? --Yair rand (talk) 00:27, 29 December 2017 (UTC)
I've changed that from "Born at sea" (instance of birthplace) to "birth at sea" (subclass of birth); and done likewise with death at sea (Q46998267). Andy Mabbett (Pigsonthewing); Talk to Andy; Andy's edits 10:39, 29 December 2017 (UTC)
I disagree with this solution. Have "at sea" fine, but why would we prefix with birth or death when it is already against at birth or death. There are all sorts of significant events that can take place at sea, so maybe just merge them all to the one. I would also suggest that the place will be the object on which the person was at the time, they are not usually floating on the sea. That accuracy is usually is in the source, so I wouldn't know why we would not wish to record that, ie. ship name.  — billinghurst sDrewth 07:06, 30 December 2017 (UTC)
The presumption "when it is already against at birth or death" is not always true. Andy Mabbett (Pigsonthewing); Talk to Andy; Andy's edits 09:25, 30 December 2017 (UTC)
@Pigsonthewing: I cannot grok that statement... presumption when its use against a property is not true. That could be applied to any addition.

No other location I know is specified for birth or death, it is the location, and we rely on the property to tell us. Why would we now have "instance of birthplace" -> "birth at sea" and "instance of deathplace" -> "death at sea". Simply have "at sea".  — billinghurst sDrewth 07:57, 31 December 2017 (UTC)

Further, it meant to be a location, and place of birth (P19) -> "birth at sea" is not a location, so it is a fail; and I truly doubt that it is a subclass of "childbirth" if it is a location-based item.  — billinghurst sDrewth 08:01, 31 December 2017 (UTC)
From the OP: "it should appear as a qualifier for birth-location which is usually [an Ocean]". Andy Mabbett (Pigsonthewing); Talk to Andy; Andy's edits 16:46, 31 December 2017 (UTC)
  • I am not familiar on how to implement a qualifier such as the above wdt:P19 / wdt:P361* wd:Q715269, can someone take me through the process step by step? --RAN (talk) 01:20, 31 December 2017 (UTC)

New userscript quickpresets

 
quickpresets at an item about a human person. Click on a preset to add it.

On christmas I wrote a small userscript: Quickpresets allows you to add common statements with only one click. It is meant to be customizable. You can configure it to present you the presets you use most often with as little clicks as possible. For example if you often edit items of persons it can present you gender, citizenship, occupation, etc. Look at the script-site to learn how to use and customize. Leave feedback here. -- MichaelSchoenitzer (talk) 13:57, 30 December 2017 (UTC)

I have sometimes wished for this tool, thanks! //Mippzon (talk) 12:01, 31 December 2017 (UTC)
Do you think it would be possible for it to show on items with no statements yet? So one could select instance of human, building etc? //Mippzon (talk) 12:14, 31 December 2017 (UTC)
@Mippzon: Yes, I can build that in. -- MichaelSchoenitzer (talk) 13:15, 31 December 2017 (UTC)
@Mippzon: And done. Just add it to the configuration with null as the qid of the preset. (Compare with my settings if unclear.) -- MichaelSchoenitzer (talk) 13:55, 31 December 2017 (UTC)
Wow, that was quick! Thanks a lot! Helps when fixing items without claims- Happy new year to you! //Mippzon (talk) 15:32, 31 December 2017 (UTC)

Constraints for residence (P551)

Hi! It seems as there is a constraint for residence (P551) around start and end dates, but it doesn't seem to be applied. Can someone take a look on why it is not activated properly? //Mippzon (talk) 10:10, 31 December 2017 (UTC)

Fixed, see Help:Property constraints portal/Mandatory qualifiers. —MisterSynergy (talk) 10:25, 31 December 2017 (UTC)
Thanks for help and guidance! //Mippzon (talk) 10:28, 31 December 2017 (UTC)
I removed the constraint given that it doesn't make a lot of sense to exist. When it comes to historic people it frequently happens that we know that they resided at some point in time in a given town/city/country but we don't know the start and end dates. ChristianKl ❪✉❫ 17:30, 31 December 2017 (UTC)

Leverage different from (P1889) in constraint checking

Property talk:P1889 gives two ideas for exploiting different from (P1889) in constraint checking (see "Robot and gadget jobs"). Any takers to implement them? We already got 230k items (125k pairs?) marked this way. --Vladimir Alexiev (talk) 16:12, 31 December 2017 (UTC)

Retrieved from "https://www.wikidata.org/w/index.php?title=Wikidata:Project_chat/Archive/2017/12&oldid=2030240085"