How would one use URLs with this property? Does it make sense to create an item for a web page merely to use it for sourcing a statement? Or would a better solution be to have another property identical to this one except that its data type is string? Or if a URL data type is planned to be added, is it best to wait until it's available? Silver hr (talk) 23:27, 10 March 2013 (UTC)Reply

Yes, the way we will use source is still rather unclear. Most webpages would not be acceptable items under wd:Notability, but in most cases, they would not be very good sources either. I think one possibility will be: add the website as the property value (when it is notable enough to be an item), and the exact URL as a qualifier. --Zolo (talk) 07:24, 11 March 2013 (UTC)Reply

I think there are legitimate cases for sourcing with webpages. I can show you an example. In Q6153452, there is a statement that the owner is Wyndham Worldwide. I think a source of sufficient authority for this statement is Wyndham's website, but I'm not sure how much sense it makes for it to be an item. If I'm interpreting the m:Wikidata/Data model correctly, there should be an IRI datatype which could, when implemented, be used in a source property, right? Silver hr (talk) 15:25, 11 March 2013 (UTC)Reply

Yes, it seems likely that we will need a differnt datatype for such cases, and an IRI datatype is foressen. Another possibility is that we wil not use this property at all, I really dont know. --Zolo (talk) 16:55, 11 March 2013 (UTC)Reply

A discussion about merging this property with P92 (legally estahblished by) is running there Snipre (talk) 09:31, 30 March 2013 (UTC)Reply

Today, I have started getting this error message when setting a reference 'stated in': "An error occurred while saving. Your changes could not be completed. Details Warning: Use imported from (P143) instead."

First off, I think that it would be a mistake to deprecate this property (P248) in favor of P143. As far as I can tell, discussions for the last two years have clearly indicated that there is a semantical difference between those two. There are reasons for retaining both properties.

Secondly, if P248 should be deprecated, where is the discussion around this? At least there should have been a reference to such a discussion from this Talk page.

To top it all off, on Property talk:P143 (and Wikidata:Requests_for_comment/Sourcing_requirements_for_bots) there's previously been discussions regarding deprecating P143 (not P248).

So I really see no valid reason for the error message that I am getting today. At the very least, this should have been discussed thoroughly (and if it has been, it should have been referenced here). Fred Johansen (talk) 13:02, 1 February 2015 (UTC)Reply

I have also seen the message on occasions, then spent a good hour trying to find a latest guide and failing. To me, I feel that P143 (imported) is for bots and P248 (stated) is for humans. P143 has been blindly copied, P248 has some thought behind it. Periglio (talk) 02:02, 3 February 2015 (UTC)Reply

The property clearly need the "source of declaration" label. The example used in the description is Richard III of England:

 place of death: Bosworth field
   reference:
     P248: Polydore Virgil

So because "stated in": "Polydore Virgil" makes no sense, the appropriate label is "source of information". That's valid for all languages that use "stated in" translations. I know that there were edits that tried to change that but they are of course not legit. --Ogoorcs (talk) 18:18, 12 October 2018 (UTC)Reply

This reference qualifier is supposed to be used with work items as values, not humans or organizations. Seems you’re looking for another reference claim, but please do not touch this one with 26M uses before consensus about changes has been reached. —MisterSynergy (talk) 19:55, 12 October 2018 (UTC)Reply

• For your sample, the value should probably be "Anglica Historia" not the item for its author. --- Jura 19:58, 12 October 2018 (UTC)Reply

The example above in this page report a human as 'source of information' and that seems correct to me.

Even the description explicitly says

   to be used in the references field to refer to the information source in which a claim is made; for qualifiers use P805

There are no indications whatsoever that this property should only be used for works. A property for other items would be redundant and useless. Further, there are lot of statements whose reference urls have 'source of informations' who are notable entities that with a works only restriction wouldn't have any other statements. You can see two example in the property page: references taken from microsoft website and the statement by Polydore Virgil (the fact that the statement was in 'Anglica Historia' does not exclude the case of public available statements not actually present in a 'work', like the ones on Microsoft website).

Also you have to note that it's your usage that is not the 'default' one, not mine. --Ogoorcs (talk) 20:12, 12 October 2018 (UTC)Reply

We don't allow original research. A statement by a person needs to materialize in some form to be added to Wikidata. --- Jura 20:18, 12 October 2018 (UTC)Reply

I never said that; also, in the context of software version statements, the signature of the software author and thus its official website is probably the MOST valid reference for the statement and that not constitute original research. Also statements automatically inferred from other wikidata statements are not original research, when the inference is itself referred.

In the case of a reference containing a link to a state organization (that is indeed valid), the ref could not point to a precise 'work'. Despite this, that organization is a valid 'source of information'. --Ogoorcs (talk) 20:32, 12 October 2018 (UTC)Reply

Something that is published by an organization is an „is a valid 'source of information'“. No the organization itself. --Succu (talk) 20:36, 12 October 2018 (UTC)Reply

• The property just indicates where the same statement can be found. Help:Sources attempts to explain how various forms are included and which properties to use. If there are cases that aren't covered, maybe this needs a new property. BTW, in the meantime, I made a few items about works for Polydore Vergil. Seems we didn't have any. --- Jura 20:43, 12 October 2018 (UTC)Reply

So should publisher (P123) or author (P50) and not 'source of information' be used to describe valid references that point to urls that are not works and that belong to humans or organizations?

EDIT: in the case of a state agency or university, if publisher is limited to 'book, periodicals, games and software', should author be used? --Ogoorcs (talk) 20:58, 12 October 2018 (UTC)Reply

Use the publication (Q732577) not related stuff. --Succu (talk) 21:09, 12 October 2018 (UTC)Reply

Sorry, can you rephrase? @Jura1: @Succu: @MisterSynergy:--Ogoorcs (talk) 21:27, 12 October 2018 (UTC)Reply

Can we change the <instance of> relation to <instance or subclass of>? Currently <stated in> Companies House (Q257303) throws a constraint violation because Companies House is an instance of “company register” which is a subclass of “Information” (a valid value for <stated in>). - PKM (talk) 23:50, 22 March 2019 (UTC)Reply

Just for the record, I understand this has been changed a few days later; see here. --Diegodlh (talk) 21:29, 23 September 2020 (UTC)Reply

This property has an "as references" scope constraint. However, links to SPARQL queries in this talk page seem wrong, as they seem to be treating it as a "as main value"-constrained property.

See for example the "List of values" link at the top of the Documentation section above. Line 9 of the SPARQL query is:

?item wdt:P248 ?value

Isn't that retrieving only items which are violating the constraint? That is, items which use the property as an "as main value"-scoped property? I am a Wikidata beginner, so maybe I'm getting something wrong. Thanks! --Diegodlh (talk) 21:58, 23 September 2020 (UTC)Reply

So How would one search for references where "stated in=some publication/something"? Get all references with "stated in=anything" (not "stated in=Q123"):

SELECT ?ref ?statedin
WHERE {
 ?ref prov:wasDerivedFrom ?a .
 ?a pr:P248 ?statedin
 # ?statedin wdt:P248 wd:Q123
} LIMIT 5

--2601:281:D87E:D6E0:A43:2FE4:251A:4D11 14:45, 15 January 2024 (UTC)Reply

It seems value type constraint has become pretty much useless ever since national library was added as valid value type about year ago (Special:Diff/973587822). Now several organization classes are listed, as opposed to original "work", "database" and alike as value types. I think library isn't an actual source where something is stated in, and rather the actual publication (may be available in some library) should be given as "stated in" value. Similarly, I'd expect other organizations to be linked using publisher (P123) or some other more appropriate property. 2001:7D0:81F7:B580:9587:D3CB:74EF:6AF3 18:26, 29 November 2020 (UTC)Reply

• Agree. --- Jura 10:34, 21 January 2021 (UTC)Reply

@Jura1: I'm a little surprised because of your massive editing of the property restrictions. It is primarily about this processing. Can you please explain why you did it. In any case, there was no discussion here on the discussion page. --Gymnicus (talk) 10:25, 21 January 2021 (UTC)Reply

• Help:Sources explains how this is property used in the reference/source section of statements. I don't really see how organizations could be used there and why they were added as allowed values. Maybe this was discussed somewhere. Organizations would generally be publishers or possibly authors of publications used as values. --- Jura 10:30, 21 January 2021 (UTC)Reply

A sports association, such as the International Bobsleigh and Skeleton Federation (Q547007), often provides information about athletes in its database, which is used here on Wikidata and also in Wikipedia. This information is therefore verified by the sports association. --Gymnicus (talk) 10:42, 21 January 2021 (UTC)Reply

How is this relevant in terms of Help:Sources? --- Jura 10:43, 21 January 2021 (UTC)Reply

I'd expect the value for P248 to be the item for the database you mention. International Bobsleigh and Skeleton Federation (Q547007) would probably be its publisher. --- Jura 10:59, 21 January 2021 (UTC)Reply

@Jura1: This can probably best be explained with an example. It's about giving Matthias Guggenberger's weight. This statement requires individual evidence and the weight of the athletes is often given in the IBSF database. For this reason, for example, the weight of Matthias Guggenberger (Q331430) is verified by the IBSF. Due to your editing an error occurs there. --Gymnicus (talk) 11:08, 21 January 2021 (UTC)Reply

• v
• t
• e

mass	75 Kilogramm edit 1 reference stated in International Bobsleigh and Skeleton Federation IBSF athlete ID 129811 language of work or name English retrieved 21. January 2021 add reference
		add value

• I think it's correct that this displays as an error. International Bobsleigh and Skeleton Federation (Q547007) should be used with publisher (P123) or author (P50). Help:Sources#Databases is the relevant section. --- Jura 11:13, 21 January 2021 (UTC)Reply

@Jura1: In point 3, however, it says that the error now displayed is incorrect. The International Bobsleigh and Skeleton Federation (Q547007) data object belongs to the property IBSF athlete ID (P2991). --Gymnicus (talk) 11:21, 21 January 2021 (UTC)Reply

• Point 3 currently reads "link it to the database item". Q547007#P31 has different statements. A database item still needs to be created. --- Jura 11:29, 21 January 2021 (UTC)Reply

@Jura1: No. At least in German there is talk of „Objekt zur Datenbank“ (english for "object to database"). And as you can look up on the IBSF athlete ID (P2991) property, that is the object International Bobsleigh and Skeleton Federation (Q547007). In addition, nowhere in the German version does it say that the "object to the database" must be a database. As in the case of the IBSF, the "object to the database" can in principle also be the operator. --Gymnicus (talk) 11:41, 21 January 2021 (UTC)Reply

It's a known problem that P2991 values are suboptimalproblematic. This is why we have issued by (P2378) and applicable 'stated in' value (P9073). Q547007 is perfect as value for Property:P2991#P2378.
However none of this changes the general idea of Help:Sources. Did you note point 2? "Ensure that at least the following statement is added to the database item: instance of (P31) → database (Q8513)". --- Jura 11:59, 21 January 2021 (UTC)Reply

It's fairly easy to fix this by bot. If you create an item for the database, you can add a request at Wikidata:Bot requests. --- Jura 12:02, 21 January 2021 (UTC)Reply

In principle, that's not my job, it's yours. – I understand the intention behind the plan. But you should have thought better of how to implement the plan. Simply doing extensive editing and then hoping that the others will correct the mistakes is definitely not the right approach. You could have looked at them beforehand, where errors could occur and eradicated them in advance. --Gymnicus (talk) 12:15, 21 January 2021 (UTC)Reply

I'm not saying it's your "job", it would just be nice if you would create an item for the database, so we could have them corrected eventually. While constraints have helped quite a lot getting statements right, one can't rely on them exclusively to ensure that additions are optimal. Constraint violation reports also help identify problems and correct them in bulk. Obviously, this can take some time, but generally eventually happens (BTW I don't think I added any of the incorrect statements).
Are there any open points about Help:Sources#Databases that need clarification? --- Jura 12:31, 21 January 2021 (UTC)Reply

@Jura1: I still believe that your approach was incorrect. Just because you and the IP (in the discussion about it) were not satisfied with the current variant, you can't simply remove it without discussion, especially when so many errors occur. You are now saying that they can now be easily removed using bots. That is certainly true. But before you could have found out by means of SPARQL queries which database properties do not have an associated database object. Then you would have had the opportunity to create a database object for each of these properties. And then you would have made a bot request recommended by you and many errors had been corrected in advance. I would call that acting with foresight and that is very important in such a far-reaching process as yours. A little tip from me for you. --Gymnicus (talk) 18:22, 21 January 2021 (UTC)Reply

Well, comment about it was made nearly 2 months before value types were restored. While in the first place and not that long ago these organization classes were added without any discussion or explanation at all, which would pretty much justify restoring the status quo even without discussion. Also, the errors are there regardless of constraint. Constraint just allows users to know that things aren't quite right, as long as we don't make the constraint useless by allowing pretty much any value. 2001:7D0:81F7:B580:A99A:4B67:BEDC:A654 11:50, 27 January 2021 (UTC)Reply

More generally, I've been wondering if it's really a good idea to display constraint warnings to all users. My impression is that average user just adds whatever class relation as long as it makes the warning go away, without caring if their "fix" actually makes sense. 2001:7D0:81F7:B580:A99A:4B67:BEDC:A654 12:00, 27 January 2021 (UTC)Reply

As there are many allowed value types, I think it makes sense for us to create some common shared parent class which captures their shared function - something like "information source". Any ideas or objections? --SilentSpike (talk) 20:10, 24 December 2021 (UTC)Reply

Most allowed classes are already subclasses of work (Q386724), thus they are redundant and could be removed immediately. The constraint classes which currently do not subclass work (Q386724) should be reviewed; if they cannot be made a subclass of work (Q386724), they might perhaps better be removed anyways. Relevant query: [1]. —MisterSynergy (talk) 20:38, 24 December 2021 (UTC)Reply

@MisterSynergy: Is work not too general? For example fictional character (Q95074) is a subclass of work, but doesn't make much sense as a source. SilentSpike (talk) 20:43, 24 December 2021 (UTC)Reply

It is not perfect, but it is effectively the gist of this constraint: we expect a *work*, not an organization, or project, or institution, or building or whatever.
I would be surprised if we could really narrow it down to a value that fits every situation, without artificially creating some class "work in the sense of P248" designed for this purpose only. —MisterSynergy (talk) 20:59, 24 December 2021 (UTC)Reply

Two candidates that come to mind are: source of information (Q3523102) or publication (Q732577). The former would make sense since (by obvious reasoning) only a source of information can possibly be usable as a source of information. The latter would perhaps be slightly more restrictive to say that only published information should be usable as a source. SilentSpike (talk) 12:53, 25 December 2021 (UTC)Reply

For some additional context, I'm reviewing various constraints and sourcing in Wikidata in an effort to make the constraints drive the data instead of driving the constraints with the data (i.e. while use of "work" makes the constraint effectively right given the current modelling, it makes more sense to me to make the constraint more restrictive so that errors in the modelling are revealed and should be fixed to meet the logical constraint). SilentSpike (talk) 12:56, 25 December 2021 (UTC)Reply

@Emu: I saw that yesterday you added news agency (Q192283) as a value type. I don't fully understand this addition as it contradicts the current handling of this property. As MisterSynergy writes in section one above, the value types are supposed to be subclasses of work (Q386724). news agency (Q192283) is that not at all. So I would like to know what you were thinking when adding it. --Gymnicus (talk) 13:43, 11 August 2022 (UTC)Reply

@Gymnicus: I did it to source Nikolaos Vernezos (Q112742832) and adding another item just for the website of Orthodoxia News Agency (Q113487152) seemed a little over the top. But feel free to undo my edit, I don’t have any strong opinion about it. --Emu (talk) 15:07, 11 August 2022 (UTC)Reply

@Emu: I took a look at your example. In this case, wouldn't editor (P98) be the better property? --Gymnicus (talk) 16:55, 11 August 2022 (UTC)Reply

@Neo-Jay: Why do you want to use this property as a qualifier as well? --Gymnicus (talk) 13:51, 11 August 2022 (UTC)Reply

I added "as qualifier" to "property scope constraint" because I think that P248 (stated in) can be used as a qualifier of P953 (full work available at URL). Is that correct? --Neo-Jay (talk) 14:09, 11 August 2022 (UTC)Reply

@Neo-Jay: Do you have an example how you would use it? So I can imagine it better. --Gymnicus (talk) 14:27, 11 August 2022 (UTC)Reply

For example, "The Taiwan Question and China's Reunification in the New Era" (Q113486427): statement "full work available at URL" with qualifier "stated in" and other 3 qualifiers. All the 4 qualifiers I added were later removed by KrBot and added to reference by it. I believe that this bot was wrong. But I don't want to spend my time on fighting a bot. --Neo-Jay (talk) 14:58, 11 August 2022 (UTC)Reply

@Neo-Jay: I think instead of stated in (P248) would be published in (P1433) the right property here and in my experience the bot wouldn't intervene either. --Gymnicus (talk) 21:11, 11 August 2022 (UTC)Reply

BBC News Online (Q4704926) is part of BBC Online (Q747860) and would be unsuitable for large parts of BBC Online. If there are no objections then I'll remove it Piecesofuk (talk) 08:08, 6 October 2022 (UTC)Reply

  You are invited to join the discussion at Wikipedia:Module talk:Wd § References mapping. {{u|Sdkb}} ^talk 20:51, 7 June 2023 (UTC)Reply