Wikipedia talk:Wikidata

Wikipedia:Reliable_sources/Noticeboard#WikiData_source

I just spotted that there's a relevant discussion going on at Wikipedia:Reliable_sources/Noticeboard#WikiData_source - but it's a bit of an echo chamber at the moment between two editors. Additional input there might be useful. Thanks. Mike Peel (talk) 14:10, 20 May 2018 (UTC)

Keeping track of what’s going on on Wikidata from Wikipedia

Hello all,

In order to answer some of the concerns that have been raised during the RfC, we've been compiling a few tools that help Wikipediens to monitor the changes happening on Wikidata. You'll find a short description of them in this blog post. I hope that is helpful, and feel free to ping me if you have questions or need precisions. Lea Lacroix (WMDE) (talk) 15:41, 1 June 2018 (UTC)

@Lea Lacroix (WMDE): Do you know if this possible to push the following Phabricator task ? This can help to reduce malicious vandalism aiming to change only part of a statement. Snipre (talk) 22:02, 1 June 2018 (UTC)

Q numbers showing up in infobox

Don't know where to place this:

• Infobox with Q numbers

? --Francis Schonken (talk) 05:42, 17 June 2018 (UTC)

The problem is that the linked items for the "has part" statements on d:Q495 don't have English labels. It looks like most of them have only Italian labels. To fix it someone could add the English labels to these items. --Lydia Pintscher (WMDE) (talk) 09:48, 17 June 2018 (UTC)

@Lydia Pintscher (WMDE): there seems to be some misunderstanding about the nature of my question. I did not ask to solve this particular instance (which, BTW, was already solved satisfactorily). The nature of my question is: what can be done to prevent similar issues in the future? Throwing WHS Wikidata infoboxes out of English Wikipedia was already decided, and that solved the issue in this particular instance. I suppose there are other Wikidata infoboxes which could show this problem, so my question is: how do we prevent that from happening in these other infoboxes (as in a structural solution)? --Francis Schonken (talk) 10:19, 17 June 2018 (UTC)

@Francis Schonken: Yes, you did ask to solve that particular instance. The solution is the same as for any issue where data is incorrect: you fix the data. More generally, these kind of rare problems can be minimised by ensuring that infoboxes are designed so that they only import sourced data by default, unless we know that every page that transcludes the infobox has been checked. That is clearly only appropriate where the number of transclusions is small enough to be curated comfortably by an individual or a WikiProject. In any case, and especially for larger scale use, we ought also to insist that the editor adding an infobox to a given article should either correct any deficiency in the data or suppress the display of unwanted data in that article by using |suppressfields=includes (or whatever field should not display).

• {{wdib |P527 |qid=Q495 |fetchwikidata=ALL |onlysourced=no}} → Crocetta Centre, , , Borgo San Donato, , , , , Lingotto,  
• {{wdib |P527 |qid=Q495 |fetchwikidata=ALL |onlysourced=yes}} →
• {{wdib |P527 |qid=Q495 |fetchwikidata=ALL |onlysourced=no |name=includes |suppressfields=includes}} →
• {{wdib |P527 |qid=Q495 |fetchwikidata=ALL |onlysourced=no |linked=no |lang=it}}} → 1 - Centro, Crocetta, 2 - Santa Rita, Mirafiori nord, 3 - San Paolo, Cenisia, Cit Turin, Pozzo Strada, 4 - San Donato, Parella, Campidoglio, 5 - Borgo Vittoria, Madonna di Campagna, Lanzo, Lucento, Vallette, 6 - Regio Parco, Barca, Bertolla, Barriera di Milano, Rebaudengo, Falchera, Villaretto, 7 - Aurora, Vanchiglia, Sassi, Madonna del Pilone, 8 - Borgo Po, San Salvario, Cavoretto, Lingotto, 10 - Mirafiori sud 

The tools are all there; it just remains for you to learn how to use them. --RexxS (talk) 11:25, 17 June 2018 (UTC)

I prefer the tool that was agreed upon for the WHS infobox. Seems far less complicated. Regarding what I asked or didn't ask: I think I'm responsible for my own intentions, thus I explained what my original intention was, i.e. asking for a structural solution. --Francis Schonken (talk) 11:40, 17 June 2018 (UTC)

Also, the example in my OP also showed "Q28769238", unrelated to the items for the "has part" statements (P527). So yes, a structural solution would be welcome. --Francis Schonken (talk) 11:57, 17 June 2018 (UTC)

See MOS:LISTGAP. If you think something different is less complicated, why don't you just use it, instead of complaining here about an issue that's already been solved? Whatever your intentions, the onus is on you to express them clearly. It's no use getting irritable when you fail to do that. If your problem was with the "Location" field, rather than the far more obvious "Includes" field, why didn't you say so? The solution, of course, is to apply an English label "Province of Turin" to d:Q28769238, just as Lydia explained to you. In that particular case, the infobox could have been designed to use only the "best" value for located in the administrative territorial entity (P131), which would have avoided that problem. Something to think about for infobox designers in future. --RexxS (talk) 19:00, 17 June 2018 (UTC)

My problem is with "Q numbers showing up in infobox(es)" – do you have a structural solution for that? --Francis Schonken (talk) 19:41, 17 June 2018 (UTC)

Actually, I agree that the infobox must show the English sitelink as the first choice, the English label as a distant second choice, and probably show nothing if none is there.--Ymblanter (talk) 20:08, 17 June 2018 (UTC)

@Ymblanter: There seems a reasonable agreement that the disambiguated sitelink is probably best on enwiki when a sitelink exists. Even without a sitelink, I can still create a link if there's a redirect on en-wiki with the same title as the label. Obviously the unlinked label remains the next best choice when there is no sitelink. That leaves the unusual situation where no sitelink or label exists in English. Is it worth giving the editor a hint that a value exists but not in English at present? The other possibility might be to return nothing as you suggest, but we then lose useful data that can be provided simply by adding an English label to the Wikidata entry. As English is the fallback language for all others, it also does a favour to the other projects by ensuring that there will then be a label available for everybody. I suppose we could return just a maintenance category like "Infoboxes with data available in other languages". There's a similar discussion of an "opposite" request at c:Template talk:Wikidata Infobox #Please don't display Q-IDs to the end user! What do others think? --RexxS (talk) 23:17, 17 June 2018 (UTC)

Thanks for the link to the commons discussion. To me it is rather self-evident to not show bare Q numbers in infoboxes, if it is technically avoidable. Do we need a consensus on that first? --Francis Schonken (talk) 05:25, 18 June 2018 (UTC)

We definitely must not show Q-numbers to the readers, I do not think we need consensus on that. But we might indeed show them to some categories of users - may be as an opt-in gadget or smth like this.--Ymblanter (talk) 05:51, 18 June 2018 (UTC)

Let's then first address the issue for the general readership: adding gadgets can be an objective if and when the general issue has been sorted. --Francis Schonken (talk) 06:28, 18 June 2018 (UTC)

Regarding the suggestion to use Wikidata labels as a (first) fall-back option: disagree, because of the multiplication effect of label vandalism (see Wikipedia:Wikidata/2018 Infobox RfC#Trocolandia – several approaches to solve the issue were proposed there, none implemented afaik). --Francis Schonken (talk) 06:45, 18 June 2018 (UTC)

Example 2: the "location" parameter of the second infobox on this page seems to contain a lot of cruft, including a Q number. Is it vandalism? cluelessness? good-faith excessive detail? ...seems rather complicated to sort. In this instance, the second infobox should not repeat the location info which is already contained in the first infobox (and it was sorted by embedding the second infobox in the first) – the point being that the crufty list of ten (!) location descriptors should and could have been somewhat more contained by at least excluding second-hand descriptors such as bare Q numbers. --Francis Schonken (talk) 07:38, 18 June 2018 (UTC)

In this example, for whatever reason, the infobox reads and displays all historical locations, whereas it only must display the actual location.--Ymblanter (talk) 07:44, 18 June 2018 (UTC)

Well, in fact, even if not embedded in the first infobox, the second infobox should not have location info at all: the location info should be in the main infobox – at best, it is a redundant doubling of infobox info, or worse, as was the case here, a rather detailed, questionable & undesirable content variation. --Francis Schonken (talk) 07:59, 18 June 2018 (UTC)

I would even argue that not the whole city is the World Heritage site, but a smal collection of buildings most of which ar in the historic center, therefore there must be a separate article on the WH site (similarly to the one I created about St. Petersburg), and the infobox belongs there. However, obviously none of these issues is a Wikidata problem.--Ymblanter (talk) 08:44, 18 June 2018 (UTC)

Displaying Q numbers in infoboxes is the topic of this thread. Summing it up for the second example (Q4322516):

• the Q number should not show up in a second infobox on a page (prior situation)
• the Q number should not show up in an embedded infobox (current situation)
• the Q number should not show up in a main infobox on a page dedicated to the specific WHS (possible future situation)

in sum, the Q number should never show up, not in any infobox shown to a general readership. --Francis Schonken (talk) 09:03, 18 June 2018 (UTC)

So what should show up:

1. When there is no sitelink, but a label in English exists?
2. When there is no sitelink or label in English?

I don't see any answers to those questions so far. --RexxS (talk) 21:49, 18 June 2018 (UTC)

@Francis Schonken: "Regarding the suggestion to use Wikidata labels as a (first) fall-back option: disagree, because of the multiplication effect of label vandalism (see Wikipedia:Wikidata/2018 Infobox RfC#Trocolandia – several approaches to solve the issue were proposed there, none implemented afaik)." This is why your complaints can't be taken seriously. First, there are not "several approaches to solve the issue ... proposed there". There are dozens of whines from you, and one, just one, indication from Was a bee about possibly using the sitelink to avoid the problem of vandalism of labels. If you claim otherwise, then tell me what the other "several approaches" are. Name them or retract your error. Secondly, on 9 June, I completed the modifications that used the sitelink where available instead of the label. That circumvents the problem in your complaint there. But I don't get "Thank you, Rexx, for addressing the problem I raised". I get "none implemented afaik". Well, you're welcome.

As for "Regarding the suggestion to use Wikidata labels as a (first) fall-back option: disagree, because of the multiplication effect of label vandalism". There's none. Nobody has brought forward a single case of where vandalism has affected the label where there is no sitelink. Why do you think it is sensible to throw out information from the infobox, to solve a problem that doesn't exist? --RexxS (talk) 22:15, 18 June 2018 (UTC)

I would say to q1 - the label must be shown, but we must make sure there is no vandalism. The items without a sitelink are more obscure typically attract less vandals, I guess we can negotiate indefinite protection of these items (only those which are used in infoboxes) on Wikidata, since they are not expected to be changed on the projects either. If the snswer to q1 is to show label, I would say the answer to q2 is show nothing and log the item; than someone can go through the logs and add an English label.--Ymblanter (talk) 07:00, 19 June 2018 (UTC)

But we probably need more test cases to understand what is going on. The one with Novgorod should not have been in the infobox anyway.--Ymblanter (talk) 07:02, 19 June 2018 (UTC)

Labels

FYI

 – Follow-up on Wikipedia:Wikidata/2018 Infobox RfC#Trocolandia

Proposed approaches:

1. Was a bee's idea: see Template talk:Wikidata location#Label text or sitelink text
2. MisterSynergy's idea: see phabricator ticket T196057
3. Make labels vandal-proof at Wikidata

Afaik Wikidata labels can not carry an external reference: they are added (and can be modified) by Wikidata users and/or are imported from sources such as Wikipedia. As such they can never be considered as deriving from a reliable source, per WP:USERGENERATED, WP:CIRCULAR and/or other aspects of WP:RS/WP:V.

To answer RexxS's question ("what should show up ... When there is no sitelink, but a label in English exists?"): at least, the label should not show up in the infobox – would breach Wikipedia's core content policies if it did. For clarity: under current circumstances, i.e. the technicalities of Wikidata not (yet) having the possibility of an external reference for a label, and the English Wikipedia's core content policies, where an exception is not (yet) allowed in this case. --Francis Schonken (talk) 07:46, 19 June 2018 (UTC)

You clearly don't understand what referencing means. I might request the name of Douglas Adams' wife:

• Jane Belson 

In the entry for Douglas Adams, Wikidata holds the id of his wife's name ("Q14623681") and a reference url for that fact: http://www.nndb.com/people/731/000023662/ so you can use that to verify the statement. You will see that the source contains the statement "Wife: Jane Belson (m. 24-Nov-1991, until his death, one daughter)". That is what referencing the statement involves.

If Jane Belson had an article on Wikipedia, we would use the sitelink from Jane Belson (Q14623681) to display her name. She doesn't have such an article, so we use the label which does exist in English. Either way, that is nothing to do with our core content policies, which requires verification for the statement that Jane Belson was Douglas' Adams' wife (not that Jane Belson's name was "Jane Belson") - see WP:BLUESKY. If you find an entry that is not supported by the reference, then you correct it, just as everyone else does when that happens.

Of course the label "Jane Belson" should show up in the infobox: it's her name.

Now let's see your answer to the question: "what should show up ... When there is no sitelink, but a label in English exists?" I'm not interested in hearing yet again what you think shouldn't show up – we've heard that a dozen times already – I want to know what you think should show up. --RexxS (talk) 12:48, 19 June 2018 (UTC)

• Example (3): the appearance of "Třebíč castle" in the infoboxes here and here (note: as with prior examples in this section the issue has already been resolved for English Wikipedia, i.e. by applying Template talk:Infobox World Heritage Site#Implementation of RfC – so for the sake of the example, please imagine this for a fully implemented Wikidata type infobox). I don't think "Třebíč castle" should have shown up in these examples. --Francis Schonken (talk) 16:17, 19 June 2018 (UTC)

If the issue has already been resolved, what is the point of wasting everybody's time here by whining about a non-problem? Why should anybody be interested in your faulty hypotheses? Give some real examples, or can it. --RexxS (talk) 23:38, 19 June 2018 (UTC)

@RexxS: again, could you please stop your whining? Tx. In example 3 the human intervention to remove a redundant label could easily have been avoided (by not showing the label in the first place). So, if you won't/can't implement a structural solution to prevent that happening again in other instances, please at least stop the whining. --Francis Schonken (talk) 07:05, 20 June 2018 (UTC)

Q numbers in infoboxes

To answer RexxS's other question ("what should show up ... When there is no sitelink or label in English?"): at least, the bare Q number should not show up. @Ymblanter: seems we need a consensus on this after all, although for me this is completely self-evident. So, until if an when there's a proposal to have something else show up that makes sense, for me the answer is clear: in this case the infobox should show no value, meaning, as if there is no Wikidata claim for the value. --Francis Schonken (talk) 06:51, 19 June 2018 (UTC)

Yes we need consensus. Having a "Q-number" is an obvious way of inviting someone to add an English label that then benefits all of the projects. It only takes a few moments to improve the entry and then that is available for everybody thereafter. There is very little argument for excluding information that is easy to supply if somebody knows that it needs to be done.

In fact, I'll make an offer: whenever you find an example of a bare Q-number displaying on enwiki, let me know and I'll fix it for you if you can't be bothered to do so yourself. --RexxS (talk) 13:10, 19 June 2018 (UTC)

I do not think readers have a slightest idea what a Q-number is, and showing it to the general reader audience is not a good idea.--Ymblanter (talk) 13:28, 19 June 2018 (UTC)

• My suggestion/preference instead of showing the hard QID would be "Unlabeled item" (small/italics/bold/etc. formatting optional) that is either hyperlinked to Wikidata or to have the edit pencil directly to the right of that text. "Unlabeled item" is more informative than "Q81276487365" and gives a cue to the reader, as is standard on WD IBs nowadays.   ~ Tom.Reding (talk ⋅dgaf)  14:39, 19 June 2018 (UTC)

We couldn't use small, Tom because it would make text in an infobox breach MOS:FONTSIZE. We could try italics, so I've knocked up a demo in Module:WikidataIB/sandbox. It would give results like this (taken from Turin (Q495)):

• {{#invoke:WikidataIB/sandbox |getValue |qid=Q495 |P527 |fwd=ALL |osd=no}} → Crocetta Centre, , , Borgo San Donato, , , , , Lingotto,  

Wouldn't we get the usual complaints from editors that readers will be surprised if they follow the links and find themselves on a different project? --RexxS (talk) 00:23, 20 June 2018 (UTC)

That's ok, I'm just spitballing. I see now that individual links are useful in addition to the pencil, in that they save mouse clicks and draw just a wee bit of attention. I'm assuming that, when labeled, they would be converted to plain text?

Something you might want to also do is place these pages into a hidden tracking category, that capable editors can more easily devour.   ~ Tom.Reding (talk ⋅dgaf)  00:47, 20 June 2018 (UTC)

And there will always be someone to complain. Wikidata is more or less a backbone to Wikipedia now, so not linking to it, or trying to avoid it is both difficult and counterproductive for all involved projects. I'd entertain reasonable arguments, but there are nary a one.   ~ Tom.Reding (talk ⋅dgaf)  00:56, 20 June 2018 (UTC)

Whining

@RexxS: could you please stop your whining, and think about solutions instead? Tx. --Francis Schonken (talk) 06:51, 19 June 2018 (UTC)

I spend all my time thinking about solutions. How about you stop whining and do the same? --RexxS (talk) 12:18, 19 June 2018 (UTC)