Talk:Cite Unseen

Archives

---

2019
2020
2025
2026

---

This page has archives. Sections older than 4 months may be auto-archived by ArchiverBot if there are more than 4.

Scientific articles marked as 'government' sources because of PMID/PMC

It seems quite strange to me that most scientific articles are classified as "government" sources solely because they contain links to PubMed (PMID) or PubMed Central (PMC). At the same time, they receive a good score regardless of which journal they come from or what the actual quality of the article is. In reality, most of these articles have nothing to do with the government, nor have they been evaluated in any meaningful way. Would it be possible to disable PMID/PMC analysis and stop categorizing sources on that basis? Wostr (talk) 12:17, 15 May 2026 (UTC)

Can you kindly provide a link to one example article? Best, SuperGrey (talk) 13:34, 15 May 2026 (UTC)

Sure, for example the last reference in pl:Iperyt bromowy that is an article in FEBS Letters. It has a link to PubMed http://www.ncbi.nlm.nih.gov/pubmed/9256235 which probably falls into URL: nih.gov on Cite_Unseen/sources/enNPPSG/2. We have thousands of such links in pl.wiki and every source with PMID/PMC is marked as 'government' and 'reliable'. Wostr (talk) 13:57, 15 May 2026 (UTC)

Hmmm. This seems to be a tricky problem because rule-based whitelists did not work. Let me investigate it a bit. Sorry for the late reply and thank you for your patience! SuperGrey (talk) 07:37, 18 May 2026 (UTC)

Oh, I think this problem is now Fixed. SuperGrey (talk) 07:48, 18 May 2026 (UTC)

This is what I did: I added ncbi.nlm.nih.gov/pubmed to the exclude whitelist of those rules: Cite Unseen/sources/government#government, Cite Unseen/sources/enNPPSG/2#enNppsgGenerallyReliable, and Cite Unseen/sources/enVSAFES#Reliable. Welcome to report more of such problems when you recognize them. SuperGrey (talk) 07:49, 18 May 2026 (UTC)

Adding an LCCN incorrectly marks sources as government sources

I've noticed that when an LCCN identifier is added to a citation template, Cite Unseen marks it as a government source, likely because the template links to lccn.loc.gov. The Library of Congress is just a library, and is not associated with any books that have an LCCN outside of owning a copy, and assigning those books an LCCN. An example of where you can see this is on en:Park, Washington#Sources. Only the two sources with an LCCN are marked as government. If it's possible to exclude lccn.loc.gov specifically from marking sources as government that'd be great. Feedmepaperr (talk) 02:58, 18 May 2026 (UTC)

Thank you for the feedback. This is a similar case to the one above. I added lccn.loc.gov to the exclude whitelist of the gov rule in Cite Unseen/sources/government#government, and the problem should be Fixed by now. SuperGrey (talk) 08:03, 18 May 2026 (UTC)

Archive.org

Is it possible to exclude the main archive.org URL if archive.org/details/muzikmagazine is on Cite_Unseen/sources/enAS? Reference with https://archive.org/details/The_War_Gases_Chemistry_and_Analysis_Sartori URL is marked as reliable based on this and it shouldn't. Wostr (talk) 02:04, 25 May 2026 (UTC)

Thanks for the feedback. I have replaced the rule criterion with "URL String" instead of "URL" for archive.org/details/muzikmagazine. SuperGrey (talk) 02:18, 25 May 2026 (UTC)

Support for grammatical pluralization in localized strings

Tracked in GitLab:
merge requests/77

The tool currently displays stats at the top of the reference list (e.g., "total X citations, Y unknown links"). However, the localized strings do not account for linguistic pluralization rules.

For instance, in the Polish localization, a hardcoded singular/plural split (or complete lack of pluralization) results in ungrammatical outputs such as "5 książki" instead of "5 książek" (5 books) or "7 cytowania" instead of "7 cytowań" (7 citations). Polish (along with many other Slavic or complex language groups) requires a multi-form plural structure based on the cardinal number (https://www.unicode.org/cldr/charts/48/supplemental/language_plural_rules.html):

• Form 1 (Ends in 1): 1 cytowanie
• Form 2 (Ends in 2-4, except 12-14): 2, 3, 4 cytowania
• Form 3 (Ends in 5-9, 0, or 11-14): 5, 7, 12 cytowań

Could we implement a proper pluralization mechanism for localization strings? MediaWiki handles this natively via {{PLURAL:}} in wikitext, or we could handle it via a native JavaScript plural rules mapping (like Intl.PluralRules) within the gadget initialization code to support more complex languages properly. Wcalenieja (talk) 21:11, 6 June 2026 (UTC)

@Wcalenieja: Thank you for the suggestion (and that's quite the plural structure!). We'll do it all in our code via Intl.PluralRules. I've started working on this, and we should have an update ready soon. ~SuperHamster Talk Contribs 06:38, 8 June 2026 (UTC)

Thank you so much :D Wcalenieja (talk) 09:57, 8 June 2026 (UTC)

Hey @SuperHamster, I have a quick question: have these changes been implemented yet? If so, there are still some strings that need to be pluralized in certain languages, including Polish, such as: all category names (advocacy, AI-generated, AI-referred, and so on, all of them) and filters active.

Similarly, the string <locale> Wikipedia (local) (e.g., polski Wikipedia (lokalnie)) needs to be corrected. In this case, the correct form should be "polska Wikipedia" (or "polskiej Wikipedii") – a new string would need to be added to the translation where the local name is specified. Also, please change the string Local wiki settings override global settings. to (Local wiki settings override global settings.), since not all languages place a period inside the parentheses; some, such as Polish, place it after the closing parenthesis.

Additionally, there is a string concatenation issue with how the count is displayed. It looks like the system is reusing the same {{PLURAL}} block for both the total count and the "Showing X of Y citations" message. UI string concatenation breaks localization here. In Polish, the word for "citations" changes completely depending on whether it stands alone or follows the word "of" (z). To fix this, Showing X of Y citations must be a single, unified string with two variables, rather than reusing a shared plural block. It should look like this in the code: Showing $1 of $2 citations -> which allows us to translate it cleanly in Polish using the second variable ($2) for the plural logic: Wyświetlono $1 z $2 {{PLURAL:$2|cytowania|cytowań|cytowań}}

Thanks in advance :) Wcalenieja (talk) 18:57, 29 June 2026 (UTC)

Is this the biggest attack vector against active editors?

Do we know how many people are using this tool? (Not that I think it's realistic to just have this data, but of particular interest are their permissions and contribution counts) Is there any way to use this more safely (without copy pasting the entire code and inexplicably getting a syntax error in line 689)? Would this work:

mw.loader.load( '//meta.wikimedia.org/w/index.php?title=User:SuperHamster/CiteUnseen.js&oldid=30490009&action=raw&ctype=text/javascript' ); // Backlink: [[m:Cite Unseen]]

Am I just paranoid and we are all just one interface admin away from everything being thrust into chaos anyway, so none of this here really matters?

Also, if my "method" above works, should this be encouraged for all scripts one wishes to import? The gigantic red box warning you about the risks, should probably have some risk mitigation advice, no? (other than "just go ask people to read the code for you") Karsenat (she/her) (talk) 14:41, 16 June 2026 (UTC)

@Karsenat: Last I checked, around a thousand editors are using Cite Unseen (this is across Meta and the various language Wikipedias, through user script installation + via gadget on the ZH and PL Wikipedias). And yes, user scripts are...quite scary, almost comically so, given they're arbitrary JavaScript scripts with minimal guardrails, able to be created and shared by anyone. It's up to each editor to decide which scripts to install and which users to trust. Since Cite Unseen lives under my username, you basically have to trust me (+ all the interface administrators who can edit any script) to not be hacked and not turn malicious one day. All of this is not too far off of what we see with browser extensions or software development at large, with projects blindly downloading dependencies and updates to those dependencies...and of course, there have been many incidents in those spaces.

This setup isn't specific to Cite Unseen. This is how all user scripts across Wikipedia/Wikimedia work. EN Wikipedia has a most-imported scripts list and some of the most popular scripts are used by thousands of editors (so Cite Unseen is up there, but not nearly the biggest attack vector in terms of number of users).

You may have heard about the recent incident where someone with elevated permissions automatically installed a malicious user script that went haywire. There are ongoing discussions about how to make the world of user scripts more secure. I think there are some easier wins, like requiring user script maintainers with X number of users to have 2FA. I think a good future would be for a setup similar to browser extensions, where users are notified about updates to scripts, can review changelogs, approve/deny the update, etc. Or having popular user scripts somehow only be able to be updated automatically based on version control, where updates go through some sort of approval process.

Yes, loading in a specific revision does work, and is a good option. I would not call that too paranoid. But you'd be pretty unique, since most users (myself included) import scripts without specifying revisions. It's not a bad idea to encourage more users to use that approach...but I think most will opt for convenience when they can. ~SuperHamster Talk Contribs 16:11, 16 June 2026 (UTC)

Consensus-curated sources' list on french Wikipedia

Tracked in GitLab:
merge requests/83

Hello. Thanks for this tool !
I just want to let you know that our consensus-curated evaluation of sources on frWP is on this page : fr:Wikipédia:Observatoire des sources. The entries are added after discussions on the related talkpage.
If that could help... Available also to help translate if needed. — L'embellie @ 07:53, 22 June 2026 (UTC)

@L'embellie: Thanks for your kind support! I have created this page: Cite Unseen/sources/frODS. If you are willing to help translate and maintain this page, that would be great! There are some other source lists under Cite Unseen/sources for your reference. Feel free to ask anything! SuperGrey (talk) 11:56, 27 June 2026 (UTC)

Oh, thanks for your feedback. I'll start updating the page as soon as possible. — L'embellie @ 12:07, 27 June 2026 (UTC)

Hello @SuperGrey. I've added and categorized the sources from ODS to the page. It was a bit laborious but I think it's correctly done as for now. Since we don't really have a firm "deprecated" category on frWP like @Chaotic Enby told me (sorry for another ping again^^), I've put there the sources that are explicitly stated to be avoided but are not blacklisted. I will continue to maintain the page as much as I can and look at the other source lists also. — L'embellie @ 00:13, 4 July 2026 (UTC)

Wonderfully done! I'll add this to the next published version as soon as possible. SuperGrey (talk) 06:32, 6 July 2026 (UTC)

Archiseek

Archiseek is considered reliable per en:WP:ARCHISEEK, yet missing from CiteUnseen Finnfrog99 (talk) 09:43, 22 June 2026 (UTC)

@Finnfrog99: updated, thanks! May take up to 24 hours to show up for you since the dataset is cached in your browser. ~SuperHamster Talk Contribs 22:50, 23 June 2026 (UTC)

Student pub symbol?

The student publications/university pages are on the database to my knowledge. This is the best image I could find that is already uploaded to Commons. It is in theming with the other non-reliability rating specific categorisation images. (Thanks to @Clovermoss for their assistance in adding URLs.) 11WB (talk) 01:18, 2 July 2026 (UTC)

Should probably courtesy ping @SuperHamster. (I'm not very active on Meta, so there may be a delay in response time from me.) 11WB (talk) 01:19, 2 July 2026 (UTC)

Actually, we already have decided an icon. See here: merge_requests/80. SuperGrey (talk) 17:50, 19 July 2026 (UTC)

Add class=noprint to the dashboard and icon

I suggest adding class="noprint" to the Cite Unseen dashboard and icon.

Currently, when a page is printed or saved as a PDF, these UI elements are included in the output. Since they are only useful on-screen, adding noprint would prevent them from appearing in exported PDFs. Thanks! Kcx36 (talk) 18:42, 17 July 2026 (UTC)

Great suggestion! Thank you, we will add that in the next update. SuperGrey (talk) 17:52, 19 July 2026 (UTC)

Suggestion for source list category

@SuperHamster: I think it would be good to have a "religious publication" category. For example, Jehovah's Witnesses have a bunch of affiliated publications that are published by the organization itself. People who are not from specific religious backgrounds are unlikely to recognize that these are not independent sources at a glance and that seems like a good use case for inclusion with Cite Unseen. Thoughts? Clovermoss (talk) 05:15, 19 July 2026 (UTC)

To be really clear, this is not me suggesting that otherwise reliable sources geared towards a specific religious audience should be listed (example). A list like this would specifically be for ones that are institutionally connected with a specific religion. This would mostly be websites, but would also mix other source types like magazines and books. I have lots of ideas for what could be included on a sourcelist like this, I just want permission and some logistical advice (unless you have immediate concerns). Clovermoss (talk) 05:24, 19 July 2026 (UTC)

@Clovermoss: Thanks for the suggestion! We can definitely consider it. Also letting you know I've moved this from Talk:Cite Unseen/Suggestions to this talk page, since a new category is a bigger consideration (the suggestions talk page is just for suggesting specific sites be added to existing categories). Best, ~SuperHamster Talk Contribs 07:36, 19 July 2026 (UTC)

@SuperHamster: Thanks for moving it to the right place and explaining why the difference in venue matters. Keep me updated on the outcome of the bigger consideration. I'd really like to work on this sourcelist if it gets created. Clovermoss (talk) 13:48, 19 July 2026 (UTC)

@SuperHamster: c:File:Religious Secondary Sourcing Hannah Clover.webm explains in a bit more detail why this could be important. Clovermoss (talk) 18:06, 25 July 2026 (UTC)

Updated icons as part of Codex

Tracked in Phabricator:
Task T405065

Just a heads up for folks watching, there is progress being made by Chaotic Enby (thank you!) on incorporating updated Cite Unseen icons into the Codex design sysytem. See this Phabricator ticket. If you're interested, please take a look and feel free to leave feedback (ideally in the Phabricator ticket, but can post here instead if you don't have access). Thanks, ~SuperHamster Talk Contribs 07:39, 19 July 2026 (UTC)

Thanks for the heads-up, I'll be following this thread too if anyone wants to provide feedback here! Chaotic Enby (talk) 12:44, 19 July 2026 (UTC)

Out of date images

File:Cite Unseen on Yemeni Civil War (2014–present) (Dec 2020) 2.png is from six years ago. Looking at RS/P on English wiki indicates that these designations (particularly for The Guardian) are out of date. Viriditas (talk) 23:27, 22 August 2026 (UTC)