Template talk:Cite Q

Access indicator

While we're at it @Pigsonthewing:, could you poke the bear over at Wikidata to support Template:Cite journal#Access indicator for named identifiers, i.e. |bibcode-access=free, |doi-access=free, |jstor-access=free, |ssrn-access=free, and the others so we don't have to manually set them on enwiki? Headbomb {t · c · p · b} 14:48, 23 August 2026 (UTC)

You'd need d:Wikidata:Property proposal for that; LMK if you'd like me to look over drafts.

You seem to have missed my questions, above, BTW. Andy Mabbett (Pigsonthewing); Talk to Andy; Andy's edits; 15:25, 23 August 2026 (UTC)

Sorry, but that form is completely unintelligible. It's a simple proposal, and 20 mins in I'm not even halfway in the fields and I have no idea what I'm even answering. Headbomb {t · c · p · b} 15:35, 23 August 2026 (UTC)

I'll do one for Bibcode and you can replicate for the rest. I'll need at least three examples with the relevant values, please.

You still overlooked my questions. Andy Mabbett (Pigsonthewing); Talk to Andy; Andy's edits; 15:39, 23 August 2026 (UTC)

I made this https://www.wikidata.org/wiki/Wikidata_talk:Property_proposal#free_access_parameters . If you want bibcode examples, here are some. Headbomb {t · c · p · b} 15:47, 23 August 2026 (UTC)

No, I need example Wikidata items for cited works with the relevant bibcode access values.

Still no answer to my questions. Why is that? Andy Mabbett (Pigsonthewing); Talk to Andy; Andy's edits; 15:57, 23 August 2026 (UTC)

What questions? And I don't know how to browse Wikidata or search it. Headbomb {t · c · p · b} 16:12, 23 August 2026 (UTC)

The ones immediately beneath and in reply to your post starting "The vast majority of DOIs are written in lowercase..." Andy Mabbett (Pigsonthewing); Talk to Andy; Andy's edits; 16:13, 23 August 2026 (UTC)

I don't know. Above my paygrade. Don't really care either. Cite journal and others could normalize them to lowercase too I suppose. Headbomb {t · c · p · b} 16:15, 23 August 2026 (UTC)

I wonder if we couldn't instead use online access status (P6954) as a qualifier of the various identifiers instead, and if the qualifier is not present we can assume a default value, such as free to read (Q24707952) or possibly open access (Q232932) as appropriate? For instance, I added a qualifier to the Bibcode property on the PLoS article "A higher level classification of all living organisms (Q19858624)" which we could pick up easily at the point Cite Q fetches the identifiers. As usual, it would then depend on the data in Wikidata being correct, but that is after all the point really. — Jonathanischoice (talk) 11:15, 27 August 2026 (UTC)

Yeah, but that's an abuse of bibcode access. That's just a link to the PLOS article. Bibcode-access=free should be for those articles hosted on the ADSABS database (see https://i.gyazo.com/9fafefac3342f199153e17c68823e280.png), e.g. Bibcode:1980JApA....1..189G Headbomb {t · c · p · b} 11:30, 27 August 2026 (UTC)

So, maybe the example one? "No detectable supernova remnant near the pulsar PSR 1930 + 22 (Q68550452)" — Jonathanischoice (talk) 12:08, 27 August 2026 (UTC)

Yes that would be a better example. The DOI is paywalled, but the Bibcode isn't. Headbomb {t · c · p · b} 12:15, 27 August 2026 (UTC)

I made , this time with a DOI and a JSTOR... That seems generally correct, so if that's all cool and the gang with the wikidata folks, the rest should be dealt with (mostly) automation, and importing when present Wikipedia values. Headbomb {t · c · p · b} 12:25, 27 August 2026 (UTC)

We could implement some defaults in Cite Q for when a qualifier is absent, e.g. usually JSTOR is subscription only, DOI is free to read at least the abstract, Arxiv is free to read, and so on. Jonathanischoice (talk) 13:17, 27 August 2026 (UTC)

Nothing should be assumed unless it can be guaranteed to be correct. SSRNs are usually free, but are sometimes paywalled. JSTOR are often free, but mostly paywalled. DOIs are mostly paywalled but often free.

If it's not known, don't assume. No information is better than a lie. Headbomb {t · c · p · b} 11:31, 29 August 2026 (UTC)

A rough cut of retracted article indicators

I've had an initial stab at it here (see Template:Cite Q/testcases#Retracted/replaced indicators). In the config I need to get the QID of the "retracted by" property, not its label, so if anyone knows how to do that properly, let me know. I managed this with a call to followQid. — Jonathanischoice (talk) 05:08, 24 August 2026 (UTC)

Hi @Pigsonthewing, have a look at the test cases I linked above and let me know how you think the indicator ought to look and behave; for now I chose the "alert" icon from the {{Icon}} list, and it links to the QID in the retracted by (P5824) or replaced by (P1366) properties as appropriate; examples. — Jonathanischoice (talk) 08:53, 24 August 2026 (UTC)

Those look good to me, as do the categories; what do others think?

Thank you for your work on this. Andy Mabbett (Pigsonthewing); Talk to Andy; Andy's edits; 10:10, 24 August 2026 (UTC)

Minor points:

• the text RETRACTED should use {{upper|Retracted}}; likewise for REPLACED
• I wonder whether "Retracted by"/ "Replaced by" might be better?

-- Andy Mabbett (Pigsonthewing); Talk to Andy; Andy's edits; 10:50, 24 August 2026 (UTC)

Oppose as implemented: Do not misuse cs1|2 template parameters for other than their intended purposes. The purpose of |id= is defined here.

For retractions, we have {{retracted}}. There does not appear to be a template for replacement notices.

—Trappist the monk (talk) 13:53, 24 August 2026 (UTC)

It would be more sensible to add a "replaced" option to {{Retracted}} than to fork that template. Andy Mabbett (Pigsonthewing); Talk to Andy; Andy's edits; 14:02, 24 August 2026 (UTC)

Why a REPLACED? What's been replaced? Retracted should be used as was designed. Wikidata may invoke it or a version of q like {{retracted q}}, but it should be explicitly bypassed in articles, e.g.

• {{cite q|Q56237061}} {{retracted q|Q61728466}} → Zaeem Aslam; Hifsa Shahid; Zeeshan Mehmood (October 2018). "RETRACTED: Ageing effects of perovskite solar cells under different environmental factors and electrical load conditions". Solar Energy Materials and Solar Cells. 185: 471–476. doi:10.1016/j.solmat.2018.05.055. ISSN 0927-0248. Wikidata Q56237061.

With the message being tweaked to reflect the template used is {{retracted q}}, or something like it. Headbomb {t · c · p · b} 17:03, 24 August 2026 (UTC)

Hi all - apologies, I'm new to this part of the ecosystem. I'll just work on retraction for now, and leave replaced by (P1366) for later. The point of this change is to get Cite Q to display the warning itself, automatically, using retracted by (P5824); I did not know about the existence of the {{retracted}} or {{retracted q}} templates. So it seems in order to allow for an intentional citation of a retracted source with Cite Q, we would also need a similar |retracted-intentional= added to Cite Q. Also, the retracted q {{QID}} template needs to be able to accept a QID starting with "Q" as currently it doubles it up. — Jonathanischoice (talk) 23:55, 24 August 2026 (UTC)

Repeating myself: Do not misuse cs1|2 template parameters for other than their intended purposes. The purpose of |postscript= is to set (or clear) the cs1|2 template's terminal punctuation. Note the maintenance message. The purpose of |postscript= is defined here.

—Trappist the monk (talk) 01:12, 25 August 2026 (UTC)

Well instead of me continuing to guess, can you recommend a field that I can use? Update: Oh I see, I can add it at the end like the erratum ID. — Jonathanischoice (talk) 01:41, 25 August 2026 (UTC)

I didn't because there isn't any cs1|2 parameter that can be used for this purpose. You've found the way to do retracted. Were I a generic reader, linking to the retraction doi would be better than linking to the wikidata page.

—Trappist the monk (talk) 13:21, 25 August 2026 (UTC)

Hi all, yes... I finally re-read the comments from you and @Headbomb and figured out to append the retracted template at the end. As suggested, I went ahead and more-or-less cloned {{Retracted}} into {{Retracted Q}} which has the different error message (describing how to modify Cite Q to allow intentional citations), and docs which basically point everyone at the original template for any other uses. There is probably a better way to do this - to allow for a different error message? Could we edit {{Retracted}} to take a |citeq=yes to display the different error? — Jonathanischoice (talk) 13:32, 25 August 2026 (UTC)

My preference is to keep those two as far apart as possible. Cite Q is a Royal PITA and I wish it were mass purged from enwiki.

However, as far as the direct question of retractions is concerned, the Aslam (2018) example behaves as I would expect it to. While WD has the bulk information, individual articles must give it the OK to indicate it is an intentional citation.

I don't know if WD supports erratums and expressions of concern, but you could equally deal with a {{erratum q}} based on {{erratum}} and {{expression of concern q}} based on {{expression of concern}}. Or whatever wizardry you did with {{cite Q}}. Headbomb {t · c · p · b} 15:38, 25 August 2026 (UTC)

I have the opposite opinion of Cite Q and use it everywhere including all my GA work. Which is why I'm here. If you have ideas about how to make it less of a PITA, short of purging it, then please let me know. Cite Q already displays erratum, so I added an explicit test case. — Jonathanischoice (talk) 19:33, 25 August 2026 (UTC) Jonathanischoice (talk) 19:33, 25 August 2026 (UTC)

Here's my laundry list for bare minimum usability

• Use |last-n=Smith / |first-n=John A. format, instead of |author-n=John A. Smith
  • Default to a standard author format, either spaced and dotted initials "John, W. A." or Vancouver-style "John WA".
• Use year, not date for books and journals. Use full dates for websites and magazines.
• Do not display ISSNs. Journals and Magazines should not display publishers.
• Support |article-number= properly in Wikidata.
• Support |bibcode-access=, |doi-access=, etc... properly in Wikidata.

Then it become inline with how things should be formatted ideally according to most citation guides. Headbomb {t · c · p · b} 22:01, 25 August 2026 (UTC)

Excellent, it appears we use the same launderette — Jonathanischoice (talk) 00:59, 26 August 2026 (UTC)

Okay, I've had another go this time appending a notice using {{Retracted Q}}, a clone of the existing {{Retracted}} template with a Cite Q compatible error message. It also now prints replacement notices based on replaced by (P1366) but it does not currently use a template to do so. Please have another look at the test cases and let me know what you think. — Jonathanischoice (talk) 13:39, 25 August 2026 (UTC)

My thoughts: I think we can use the same pattern from within Module:Cite Q of expanding and appending {{Retracted}}, to also use {{Replaced}} (or as a parameter of Retracted?), {{Expression of concern}} (not currently a WDP afaik but should be added), and {{Erratum}} (which is already in Cite Q but not using the template output). It would make sense to pass a parameter something like |citeq=yes to the existing ones to modify their red "intentional" error messages, rather than have several separate but otherwise identical templates. This way, we can modify Cite Q to have a single |citation-intentional=yes parameter, which covers the intent parameters for any or all such templates from within Cite Q. — Jonathanischoice (talk) 01:38, 26 August 2026 (UTC)

This diff for Module:Cite Q and this diff for Module:Cite Q/config adds the retracted warning to any Wikidata item that has a value in retracted by (P5824) by including the output of the {{Retracted}} template. This is only the retraction notice, not the other things discussed above (deferred until later). — Jonathanischoice (talk) 11:11, 29 August 2026 (UTC)

Not ready yet. It is poor practice to fork {{Retracted}} → {{Retracted Q}}. Modify {{Retracted}} so that it accepts a control parameter that causes it to emit the {{cite q}}-specific message.

We should not require readers to go to wikidata to find the source of the retraction notice. Such notices often have a doi or other identifier; use that identifier instead of the wikidata qid.

—Trappist the monk (talk) 12:02, 29 August 2026 (UTC)

I'll make the change to {{Retracted}}. I'm happy to add the same logic as the retracted template to add whichever identifiers are on the WD item, but this being Cite Q, we should retain the QID of the retraction. — Jonathanischoice (talk) 12:15, 29 August 2026 (UTC)

This is nearly done, I will update the diff in the request above once it passes any existing bibcode, DOI, PMC or PMID to the original Retracted template, now modified to take |citeq=yes for the modified message. Jonathanischoice (talk) 13:12, 29 August 2026 (UTC)

Apologies for the haste; this is now ready. Jonathanischoice (talk) 13:28, 29 August 2026 (UTC)

How would it handle |bibcode-access=free/|doi-access=free/|id=? Presumably there would be a need for something like |retracted-bibcode-access=/|retracted-doi-access=/|retracted-id= or something like it? Headbomb {t · c · p · b} 13:56, 29 August 2026 (UTC)

I don't know, because we haven't implemented that anywhere yet. Let's do one thing at a time, we can burn that bridge when we get to it later. — Jonathanischoice (talk) 14:28, 29 August 2026 (UTC)

Also, that is literally the next thing on the list, but I think it would be good to get these changes out of sandbox first, before it starts to drift too much and get too difficult to manage. Jonathanischoice (talk) 21:35, 30 August 2026 (UTC)

@Trappist the monk: Does the latest change satisfy your concerns? Andy Mabbett (Pigsonthewing); Talk to Andy; Andy's edits; 12:24, 31 August 2026 (UTC)

This latest round of changes to the module arose because there are those who believe that it bangs too hard on wikidata. Do we really need the call to wikidata at line 676? If opt_cat were declared and initialized just before the retracted test (lines 618–636 then opt_cat can be modified in the test immediately after {{retracted}} is expanded. If ever 'replacement' is implemented, handle it in the same way without the call to wikidata at line 679.

—Trappist the monk (talk) 13:41, 31 August 2026 (UTC)

 Done, adjusted diff above. — Jonathanischoice (talk) 21:09, 31 August 2026 (UTC)

{{Retracted Q}} should be deleted.

—Trappist the monk (talk) 13:47, 31 August 2026 (UTC)

Tagged with {{db-g7}} Jonathanischoice (talk) 21:07, 31 August 2026 (UTC)

Is that yes, or no? Andy Mabbett (Pigsonthewing); Talk to Andy; Andy's edits; 14:00, 31 August 2026 (UTC)

I'm content for the sandboxen to be made live.

—Trappist the monk (talk) 13:45, 1 September 2026 (UTC)

 Done. Please check. Andy Mabbett (Pigsonthewing); Talk to Andy; Andy's edits; 18:43, 1 September 2026 (UTC)

@Pigsonthewing, @Headbomb - I've created a Wikidata property proposal for the similar expression of concern, which we can make behave similarly to retractions and erratum. — Jonathanischoice (talk) 04:22, 6 September 2026 (UTC)

References

Invisible testcase diffs are URL encoding in Z39.88

The diffs in the test cases are due to URL encoding differences in the OpenURL span (look for it using the span.Z3988 CSS selector) where the title has characters that need URL encoding, e.g. a single-quote or ampersand, e.g. Thom%26%2339%3Bs+Irish+Who%26%2339%3Bs+Who in the live code (left) vs. Thom%27s+Irish+Who%27s+Who on the right (sandbox). Which is odd, because there aren't any code differences I can see in the sandbox that would do that, and I've tried purging the cache etc. It's pretty minor, but noting here for the record. — Jonathanischoice (talk) 01:10, 2 September 2026 (UTC)

Some lookup efficiencies

Please see diff for Module:Cite Q and diff for Module:Cite Q/config. I've added a bit of lookup efficiency by adding a qidonly marker in the config, which indicates to the main loop to fetch the QID rather than the string label of the property. I've moved the lookups for replaced by (P1366) and corrigendum / erratum (P2507) into the config template alongside retracted by (P5824). This saves double lookups when we only need the QID. — Jonathanischoice (talk) 12:20, 2 September 2026 (UTC)

Is this included in the change request, below? Andy Mabbett (Pigsonthewing); Talk to Andy; Andy's edits; 07:20, 3 September 2026 (UTC)

Sorry, yes - I left this as a separate diff to see the changes without the erratum fix, but they are included below. — Jonathanischoice (talk) 09:31, 3 September 2026 (UTC)

 Done, Jonathanischoice (talk) 20:28, 4 September 2026 (UTC)

Make erratum use the standard Erratum template

Please see diff for Module:Cite Q and diff for Module:Cite Q/config. This is now enabled in the sandbox as for retracted notices, see test cases at Template:Cite Q/testcases#Erratum. It also builds on the qidonly lookup changes (see topic above). — Jonathanischoice (talk) 02:37, 3 September 2026 (UTC)

I'll do this tomorrow, if there are no objections. Andy Mabbett (Pigsonthewing); Talk to Andy; Andy's edits; 07:20, 3 September 2026 (UTC)

I see you've just been granted TE right—congratulations!—so I'll leave it to you to do, once people have had time to comment. Andy Mabbett (Pigsonthewing); Talk to Andy; Andy's edits; 17:47, 3 September 2026 (UTC)

 Done, Jonathanischoice (talk) 20:28, 4 September 2026 (UTC)

Disable CiteSeerX

I've disabled it for now in the sandbox by unsetting the property before rendering, and commenting it out elsewhere (just in case Penn State find some money behind one of the couches and reinstate it); see test cases. — Jonathanischoice (talk) 23:36, 3 September 2026 (UTC)

 Done, Jonathanischoice (talk) 20:29, 4 September 2026 (UTC)

Don't display ISSNs

I agree with @Headbomb that we shouldn't be displaying ISSNs for any Wikidata item that isn't an actual instance of (P31) journal, book series or other periodical, which would be an odd thing to cite anyway. Identifiers for articles, papers, chapters etc. should be to the item itself, not the item within which it is contained or published. Disabling issn in the sandbox config works, unless someone has added a value for ISSN (P236) to the WD article item. Thoughts? — Jonathanischoice (talk) 23:55, 3 September 2026 (UTC)

Publisher and publication date tweaks

I've tweaked the sandbox to: 1. suppress |publisher= for journal articles; 2. Use only the year for |publication-date= in journal articles and books; and 3. only use {{Cite book}} if |isbn= is set and |journal= is not set (some journal articles have ISBNs, which was throwing it to cite book). These changes were suggested by @Headbomb above. I've noticed in the test cases that it is not distinguishing between news articles and journal articles, so I need to figure out a good way to do that from the WD properties; both use published in (P1433) but news articles seem to be distinguished mainly by being an instance of (P31) news article (Q5707594). If anyone has bright ideas, then I'm All Ears (Q55614156) — Jonathanischoice (talk) 06:14, 6 September 2026 (UTC)

Are you sure that's right? Don't journal citations commonly use month year dates for the issue that holds the cited article?

—Trappist the monk (talk) 12:59, 6 September 2026 (UTC)

The vast majority of journals are cited with year, volume, page/article-ID only. Issue is often omitted, sometimes present, but it's at least useful. Month is rather irrelevant, bibliographically speaking, and nearly redundant with issue. See, e.g.

• APS journals (J. Götzfried, A. Döpp, M. F. Gilljohann, F. M. Foerster, H. Ding, S. Schindler, G. Schilling, A. Buck, L. Veisz, and S. Karsch, Phys. Rev. X 10, 041015 (2020).)
• IEEE journals (S. Asiri, Y. Xiao, S. Alzahrani, S. Li, and T. Li, “A survey of intelligent detection designs of HTML URL phishing attacks,” IEEE Access, vol. 11, pp. 6421–6443, 2023.)
• ACS journals (Wu, Y.; Xu, M.; Liu, S. Generative Artificial Intelligence: A New Engine for Advancing Environmental Science and Engineering. Environ. Sci. Technol. 2024, 58 (40), 17524−17528.)
• Frontiers journals (Miller, N. R., Bamber, J. C., and ter Haar, G. R. (2004). Imaging of temperature- induced echo strain: preliminary in vitro study to assess feasibility for guiding focused ultrasound surgery. Ultrasound Med. Biol. 30, 345–356. doi: 10.1016/j. ultrasmedbio.2003.11.016)
• Joule Inc/CMA journal (Purkey E, MacKenzie M. Experience of healthcare among the homeless and vulnerably housed a qualitative study: opportunities for equity-oriented health care. Int J Equity Health 2019;18:101.)

I'm not saying month is never used to cite a journal, but it's an extreme minority of cases when it is, and likely from a newspaper when reporting something like "In the 8th December edition of this year, the Journal of Foo, scientist announced..." because they want to anchor the story around a certain time for reader interest. In the footnotes, it'll still often be Smith J. (2020). Journal of Foo 42 23–45. doi:10.5555/.... Headbomb {t · c · p · b} 19:13, 6 September 2026 (UTC)

At this point I'm beginning to wonder why the template does about 50 separate property queries from the array in config, and then more statement queries to fetch authors and editors, when one query to getEntity would get us everything in one go. Even though it's an expensive query, it is probably more efficient to only look everything up once, and would give us all the claims with all the extra metadata we need for O(1) future changes i.e. without having to do further queries — Jonathanischoice (talk) 21:16, 6 September 2026 (UTC)

Okay... some more progress today. So far, the sandbox diff has the following changes:

• Use {{Cite news}} based on instance of (P31) = news article (Q5707594) or magazine article (Q30070590) (other plausible items can be added); this alone has tidied up a few test cases that were previously using {{Cite journal}} based solely on published in (P1433), and breaking some params.
• If using {{Cite news}}, some WD items are using publisher (P123) instead of published in (P1433) for the periodical, so move the value from |publisher= to |journal=.
• Suppress |publisher= when using {{Cite journal}} only if there is no ISBN - articles can be published as chapters in books and conference proceedings. This has improved several test cases, particularly articles with an ISBN that were using the book template and complaining about an unknown |journal= parameter (these now use the journal template; see below).
• Use {{Cite book}} when there is an ISBN, but only if there is no value for published in (P1433); books in a set or series should be using part of the series (P179). This has left the book test cases unchanged, but improved many of the journal article cases (since they were incorrectly being shoe-horned into the book template).
• Truncate the publication date to the year for {{Cite journal}} and {{Cite book}}, but not for news items.

Please have a look at these changes and let me know what you think. It might be worthwhile adding some more test cases where some of these conditions are not met; and possibly, separate the test cases into a small number of smaller pages, it's getting rather long and heavy to render. — Jonathanischoice (talk) 02:35, 7 September 2026 (UTC)

Handle expressions of concern

First up, I've put up a Wikidata property proposal at d:Wikidata:Property proposal/has expression of concern, which will allow us to automatically show notices using {{Expression of concern}} the same way we do for errata and retractions; feel free to comment on there to get that moving. Once that's in, it will be a simple matter to add it at the Cite Q end. — Jonathanischoice (talk) 00:21, 7 September 2026 (UTC)