Talk:Community Wishlist/Wishes/Automatic citation tool

This page is for discussions related to the Community Wishlist/Wishes/Automatic citation tool page.   Please remember to: Sign posts using the four tildes (~~~~) Remain civil and polite during discussions. Place new text under old text (start a new post).

Thank you!

Hey @Superb Owl - first off, I love your username. Second, I think this wish is interesting, but it's unclear what sorts of improvements you're seeking. Could you add some of the challenges you face, or opportunities you see via the Automatic Citation Tool? JWheeler-WMF (talk) 17:48, 26 July 2024 (UTC)

Sure! I can generate a table of perennial websites with the attributes that are missing if that would be helpful or can just give a few examples Superb Owl (talk) 18:02, 26 July 2024 (UTC)

Sure, feel free to share a table, and offer a few ideas of what's missing, or how it could be improved? Sharing the current workaround or workflow can help us better understand the current gaps in experience. JWheeler-WMF (talk) 18:04, 26 July 2024 (UTC)

I am currently filling-out a table in my community wishlist sandbox starting with the sites that seem to have the most room for improvement Superb Owl (talk) 18:14, 26 July 2024 (UTC)

Love it! JWheeler-WMF (talk) 19:14, 26 July 2024 (UTC)

Thanks for asking follow-up questions! And feel free to use the sandbox for anything and everything that is helpful Superb Owl (talk) 19:42, 26 July 2024 (UTC)

Also - someone shared with me that this tool may help solve your problem: Web2Cit JWheeler-WMF (talk) 19:54, 26 July 2024 (UTC)

I did not know about this - thanks for passing along. After testing with a few websites, I seem to get the exact same results just with two suggested citations instead of one. I also generally avoid installing user scripts for security reasons. Superb Owl (talk) 20:38, 26 July 2024 (UTC)

@JWheeler-WMF, do you know anyone that works on Citoid that can help us figure out what our next steps should be? Superb Owl (talk) 16:34, 28 July 2024 (UTC)

Great initiative Alenoach (talk) 20:03, 26 July 2024 (UTC)

There are also other issues with the tool such as that it writes staff writer into author name fields when such are common for a website and could be filtered out if even an improved author name detection code doesn't fetch the name properly. Also I think reFill and the citation bot get more / better info from the sources so maybe some code there should be added to the citation tool which would also reduce the clutter on watchlists. For the proposal I think more info on what (how it) should and could be improved would be good. Prototyperspective (talk) 22:11, 26 July 2024 (UTC)

I agree on common mistakes like 'Staff' or 'ABC NEWS' that make it into the author name - would be great if they could be filtered out.
It looks like reFill is based on the same software as the Automatic Citation Tool citoid which primarily uses Zotero (here is the library of Zotero 'translators') Superb Owl (talk) 04:07, 28 July 2024 (UTC)

Are you sure this applies to reFill? Couldn't find this mentioned in its GitHub repo. So I think the first step would be making the integrated citation tool as good as the latest version of reFill, then as next step improving on it, for example via users registering whenever there are flaws somehow so at least the most common ones are solved plus adding more info that it fails to retrieve through systematic checks like the table you created for some major news-sites. Prototyperspective (talk) 10:19, 28 July 2024 (UTC)

According to this documentation, Citoid (which uses Zotero) is the main parser Superb Owl (talk) 14:22, 28 July 2024 (UTC)

Is Citoid or reFill or something else used when clicking on the "Autofill" magnifier icon when editing a WP article in the source editor? (Under Cite -> Templates -> e.g. cite news). That's where my confusion was from partly – the citoid page only says provide use of the citoid service to VisualEditor and next to the Autofill button there is no info on which tool is used (and maybe some small info should be added there as a tooltip). Prototyperspective (talk) 17:34, 28 July 2024 (UTC)

ReFill definitely uses Citiod, there's even a file on it in the repo. Robertsky (talk) 12:22, 6 August 2024 (UTC)

See also w:WP:AGCR for more issues with automatically-generated citations (regardless of tool). * Pppery * _{it has begun} 21:40, 2 August 2024 (UTC)

Good to see that this project has been informed about this proposal. I think it may be good to add info from the reply or a link to it to the proposal somehow, maybe to here or to an eventual phabricator issue. Prototyperspective (talk) 23:30, 31 August 2024 (UTC)

done Superb Owl (talk) 16:06, 12 November 2024 (UTC)

Zotero

@JWheeler-WMF how often/frequently is WMF's Zotero translators repo updated? If I am right, the last time the update was in April 2024 according to Gitlab ([1])? In the interest of timeliness, any plans to make the updates more regular or automated? (I largely lost the motivation/interest to update Zotero translators I created because of the irregular updates from upstream... and also irregular updates with the upstream repo). Robertsky (talk) 12:28, 6 August 2024 (UTC)

I'd really like to push for more usage (and integration with Citoid) of Web2Cit. All translators are community-configurable and go live immediately. I've added quite a few for most popular Croatian (hr) portals here. And yes, it really sucks that publishers don't care about their metadata. pon_or (talk) 03:46, 7 October 2024 (UTC)

So far Web2Cit does not seem to have been trained on Reuters, NYTimes, NPR (or many sites in general) that are the biggest priority but would be great to have something that goes live when we publish, especially since some of the sites might be changing their code to try and block bots and Web2Cit and Zotero might be getting caught in the crossfire... Superb Owl (talk) 16:05, 12 November 2024 (UTC)