Wikidata:Requests for comment/Referencing websites: how do we store the website title?
An editor has requested the community to provide input on "Referencing websites: how do we store the website title?" via the Requests for comment (RFC) process. This is the discussion page regarding the issue.
If you have an opinion regarding this issue, feel free to comment below. Thank you! |
The following discussion is closed. Please do not modify it. Subsequent comments should be made in a new section. A summary of the conclusions reached follows.
- The conclusion is that published in (P1433) should be used wherever possible. If the website is notable, then an item should be created for it. If not use title of broader work (P6333). Created title of broader work (P6333), updated Help:Sources#Web_page.--Micru (talk) 13:10, 4 January 2019 (UTC)[reply]
Help:Sources does not specify how to reference the website that a web page is published on. All standard citation styles require this field when citing websites. For example, the correct way to cite a website is in MLA style is:
“Title of Web Page.” Title of Website, Publisher, Date published in Day Month Year format, URL.
“Restricting the World.” Wikimedia Deutschland Blog, Wikimedia Deutschland, 22 February 2013, https://blog.wikimedia.de/2013/02/22/restricting-the-world/.
This is equivalent to the "website" parameter in the Template:Cite web template on en wiki:
{{cite web |url= |title= |last= |first= |date= |website= |publisher= |access-date= |quote=}}
{{cite web |url=https://blog.wikimedia.de/2013/02/22/restricting-the-world/ |title=Restricting the World |last=Vrandečić |first=Denny |date=2013-02-22 |website=Wikimedia Deutschland Blog |publisher=Wikimedia Deutschland |access-date=2018-11-29 |quote=One of the features ... of Wikidata is that you can choose any item as the value for a property.}}
We need the ability to add the statement pointing to the value for "Title of Website." The only current way to this is to use published in (P1433) and create a new item for the website itself. I have created a property proposal to store this information as a string, however, despite having a lot of discussion, there is no consensus, hence the need for this RFC.
We need to develop a consensus on how to add this information, either create or not create the property depending on its outcome, and then update Help:Sources to reflect the result of this RFC.
Here are the options:
Contents
- 1 Options
- 1.1 Create a new item for any website that is used in a reference and use published in (P1433)
- 1.2 Create new property to store website title as text in the reference, but recommend using published in (P1433) if item exists or is notable
- 1.3 Create new property to store website title as text in the reference, but convert to using published in (P1433) once in a while
- 2 Comments on options
Options[edit]
Create a new item for any website that is used in a reference and use published in (P1433)[edit]
- Pros: Users are able to click on item to get more information about website
- Cons: Will result in websites that *may* not be desirable being added as items, i.e. blogspot subdomains
Votes[edit]
- Support Andy Mabbett (Pigsonthewing); Talk to Andy; Andy's edits 12:09, 27 November 2018 (UTC)[reply]
- I am most sympathetic to this option, but we need to talk specifics--the con for example. Why should we care about subdomains for the first pass of whatever tool is being created? Let's start with the easy part and just work from domain. That should save us some time and energy. --Izno (talk) 13:56, 27 November 2018 (UTC)[reply]
- @Izno: The subdomain and the domain are often different websites with different authors and publishers associated with it. i.e. obit.mlive.com has publisher Legacy.com but mlive.com has publisher Advance Local Media LLC. Or sometimes the domain points to the publisher. Mvolz (WMF) (talk) 23:11, 29 November 2018 (UTC)[reply]
- Which is fine. I don't want to create random Blogspot sources (not that we should be sourcing anything to Blogspot probably). For something like mlive.com, the subdomain is still ultimately published by, and exists at the luxury of, the publisher of mlive.com. We can refine things down the road if we think it's necessary for a particular website just by normal wiki editing (but color me skeptical). Let's avoid the pain and misery of dealing with the premature subdomains for right now. --Izno (talk) 04:11, 30 November 2018 (UTC)[reply]
- @Izno: The subdomain and the domain are often different websites with different authors and publishers associated with it. i.e. obit.mlive.com has publisher Legacy.com but mlive.com has publisher Advance Local Media LLC. Or sometimes the domain points to the publisher. Mvolz (WMF) (talk) 23:11, 29 November 2018 (UTC)[reply]
- Comment If the website has an Item, should we not be using <stated in> [WEBSITE] qualified with <reference URL> rather than just <reference URL>? - PKM (talk) 21:36, 27 November 2018 (UTC)[reply]
- On the property proposal, I originally proposed to use "stated in" but the comments there were in favour of "published in," as stated in implies that the data can be found on the url to the website (i.e. Example.com as opposed to http://www.example.com/titleofarticle Mvolz (WMF) (talk) 16:30, 28 November 2018 (UTC)[reply]
- Support The example noted as undesirable is fine. Blogspot subdomains are like imprints of a publisher. Blue Rasberry (talk) 18:10, 28 November 2018 (UTC)[reply]
- Weak support the least worst option. Cheers, VIGNERON (talk) 19:06, 28 November 2018 (UTC)[reply]
- Support with the caveat that a website title/"published in" link is not needed in all cases (see discussion about organizations below). ArthurPSmith (talk) 21:21, 29 November 2018 (UTC)[reply]
- Support --- Jura 21:39, 29 November 2018 (UTC)[reply]
- Support; better for structuring and analysis, and for handling e.g. a website with a common name which has used multiple domains, or a domain that has been used by multiple separate websites with the same name. Jc86035 (talk) 16:12, 19 December 2018 (UTC)[reply]
Create new property to store website title as text in the reference, but recommend using published in (P1433) if item exists or is notable[edit]
- Pros: Makes it easy for automated tools to add website title without having to create new item, and avoids creation of potentially non-notable websites
- Cons: Users won't be able to click on the title to get more information about the website itself.
Votes[edit]
- Oppose. This is not structured data. Andy Mabbett (Pigsonthewing); Talk to Andy; Andy's edits 12:10, 27 November 2018 (UTC)[reply]
- Support. - PKM (talk) 21:31, 27 November 2018 (UTC)[reply]
- Support also makes sure than changes in website title do not change the reference. -- JakobVoss (talk) 08:00, 30 November 2018 (UTC)[reply]
Create new property to store website title as text in the reference, but convert to using published in (P1433) once in a while[edit]
Votes[edit]
- Support --- Jura 21:39, 29 November 2018 (UTC)[reply]
- Support useful to find out if website turns out to be used more often for references--- JakobVoss (talk) 08:01, 30 November 2018 (UTC)[reply]
Comments on options[edit]
Please comment here if you think neither of these options are suitable for you and what you'd prefer instead, or if you have different pros/cons than what is listed above.
I’ve always used title (P1476) in the reference for this, so I’m confused why this isn’t mentioned as an option? Or does that mean something different? I’m not alone in using that property, at least – according to this query, 2415876 statements (1047758 distinct references) use that property. --Lucas Werkmeister (talk) 12:37, 27 November 2018 (UTC)[reply]- Disregard that, I didn’t see the link to the property proposal and missed that this is about the “container” title, not the individual page title. (May I suggest rewording the RfC to make that distinction clearer?) --Lucas Werkmeister (talk) 12:38, 27 November 2018 (UTC)[reply]
Thanks, do you have a suggestion to do so? I knew it was going to be confusing which is why I highlighted the equivalent bits in green... not quite sure how to make it clearer. Feel free to edit! Mvolz (WMF) (talk) 16:30, 28 November 2018 (UTC)
- @Mvolz (WMF): Yeah, looking at it now I really have no idea how I managed to miss that… but I’ve added an example now, perhaps that helps. --Lucas Werkmeister (talk) 14:32, 29 November 2018 (UTC)[reply]
- I can’t decide which option I should support. References typically contain stable information about the source which was valid at the time of reference creation, hence we do not update it later (e.g. if the title of the website changes, an older reference should not be updated with the new title). This approach would favor creation of a new string property that holds the title valid when the reference was created; working with multiple values and start time/end time qualifiers in separate items is probably something that the majority of editors does not understand sufficiently :-( That said, strings are pretty unstructured and not easy to manage. Difficult one… —MisterSynergy (talk) 14:10, 27 November 2018 (UTC)[reply]
- The change of name of the ressource is far from specific to this issue, for example it occurs in the place of birth of someone if the place changed its name but there is no reason to use a different item besides the change of name. It should eventually be handled by inferences and it can be handled by code of the data users, we started to do this in frwiki in biography infobox recently by using the official name (P1448) statements for example. author TomT0m / talk page 15:12, 27 November 2018 (UTC)[reply]
- Don't we generally use object named as (P1932) to capture the name/title of something at the time of use? ArthurPSmith (talk) 00:31, 28 November 2018 (UTC)[reply]
- In the cases I encountered in the tests, I don’t remember seeing this qualifier used. It’s also convenient in a database point of view to put this information in the main item and not in every statements in which this could be an issue. author TomT0m / talk page 10:41, 28 November 2018 (UTC)[reply]
- Don't we generally use object named as (P1932) to capture the name/title of something at the time of use? ArthurPSmith (talk) 00:31, 28 November 2018 (UTC)[reply]
- The change of name of the ressource is far from specific to this issue, for example it occurs in the place of birth of someone if the place changed its name but there is no reason to use a different item besides the change of name. It should eventually be handled by inferences and it can be handled by code of the data users, we started to do this in frwiki in biography infobox recently by using the official name (P1448) statements for example. author TomT0m / talk page 15:12, 27 November 2018 (UTC)[reply]
- Comment How do we handle cases where the publishing website is the official website of some organization? We probably don't want separate items for "XXX Organization" and "XXX Organization's official website", right? ArthurPSmith (talk) 17:48, 29 November 2018 (UTC)[reply]
- publisher (P123) should do the trick. author TomT0m / talk page 18:36, 29 November 2018 (UTC)[reply]
- I have no idea :). But it's a good point. In some cases the published in and the publisher would point to the same item perhaps? Or else, as you point out, they'd be split. Mvolz (WMF) (talk) 23:11, 29 November 2018 (UTC)[reply]
- Websites (any works, not just websites) on en.wp are treated as works rather than used as their publishers where they share a name (or similar name). (Though we do have items on e.g. Amazon.com and Amazon IIRC and Google.com versus Google, this is the uncommon, super-popular, few.) --Izno (talk) 04:15, 30 November 2018 (UTC)[reply]