This is The Earwig's talk page, where you can send him messages and comments.

Put new text under old text. Click here to start a new topic. New to Wikipedia? Welcome! Learn to edit; get help. Assume good faith Be polite and avoid personal attacks Be welcoming to newcomers Seek dispute resolution if needed

Archives: 1, 2, 3, 4, 5, 6, 7, 8, 9, 10, 11, 12, 13, 14, 15, 16, 17, 18Auto-archiving period: 60 days

Hello. I noticed that large chunks of this section of herbicide are copied directly from this source_{(you'll need to log in)} but the copyvio detector doesn't pick it up: [1]. I can't find a tool to show it nicely, but it is especially obvious if you look at the original diff: [2]. Presumably it isn't detected because the tool can't access the full text? I just wondered whether you'd considered linking up the detector with WP:TWL so that it can check the full text? Admittedly, I am not sure whether the publishers permit automated access, but you would think that they would like us to be checking whether their copyright is being violated! @Samwalton9 (WMF): just in case they can add anything. SmartSE (talk) 10:29, 19 December 2023 (UTC)[reply]

@Smartse It's an interesting idea! I don't think we could do anything immediately, but if it would be feasible/helpful we could initiate a conversation with one of more of the library's partners about this. Perhaps EBSCO, given that they're our search provider? I'm not sure on the details of how this would work. Samwalton9 (WMF) (talk) 12:56, 19 December 2023 (UTC)[reply]

Hey Smartse. I'm with Samwalton9 that this would be really cool to support, but I'd be very surprised if TWL's partners would be willing to open up a service to us that would enable the copyvio detector to check content programmatically. Initiating a conversation couldn't hurt, though. — The Earwig (talk) 03:56, 21 December 2023 (UTC)[reply]

@The Earwig It's not impossible to imagine - TWL's partners are often concerned that WP editors are going to be copying content, so being able to say "we want to make absolutely sure that's not happening" could be seen quite positively. Would EBSCO be the right organisation, do you think, since they run (and provide us with) EBSCO Discovery Service? Samwalton9 (WMF) (talk) 09:51, 21 December 2023 (UTC)[reply]

@Samwalton9 (WMF): I was initially thinking of just searching the sources cited in the article. Apparently, most of the full texts can be accessed by appending the DOI to https://doi-org.wikipedialibrary.idm.oclc.org/ so it shouldn't be too difficult to programmatically access the full text (not withstanding the authentication and any rate-limiting) and then the text could be compared as the tool already does. I'm not familar with EBSCO, but I imagine that using that would be more complicated as you would need to take chunks of the article, query the search engine repeatedly and then check full texts that could be matches. I also posted about this at meta:Talk:CopyPatrol#Can_the_tool_access_paywalled_full_texts? and the ithenticate service can detect it in a new edit - see the hit for link.springer.com - even though the full text is paywalled, so maybe using that service in this tool could be an option as well? It seems like that tool does a pretty good job of catching new copyvios but we are less capable of detecting old instances. SmartSE (talk) 12:26, 21 December 2023 (UTC)[reply]

Checking the DOIs of sources directly cited would be a good start and wouldn't require us to get a search engine working, so we could try that (though the full scope is of course somewhat limited). If I'm to do that through TWL's proxy, we'd need to get the bot access somehow and confirm this usage is within their terms. @Samwalton9: I'm also unfamiliar with EBSCO and from skimming the linked pages it's not clear to me if they offer a search API that I would be able to use for what SmartSE described (query the search engine repeatedly given text snippets from the article and receive results that enable me to get the full text of the source for comparison). I see discussion of end-user search tools, but not an API. One change to the copyvio detector I am sure we will need to make is not showing the user the full text of the suspected source, only the copied snippets. — The Earwig (talk) 14:19, 21 December 2023 (UTC)[reply]

@The Earwig Is this a helpful link? Once we've confirmed this is a viable and useful approach I'd be happy to bring this up with them. Samwalton9 (WMF) (talk) 16:07, 8 January 2024 (UTC)[reply]

@Samwalton9 (WMF): Probably. I can't say for sure (the API documentation requires an account, and I still don't know the terms of use), but it looks like the right direction. Thanks! — The Earwig (talk) 17:01, 8 January 2024 (UTC)[reply]

Alright, I'll get an initial conversation kicked off with them and see how feasible this is. I'll be in touch! Samwalton9 (WMF) (talk) 10:33, 12 January 2024 (UTC)[reply]

@The Earwig Good news! We met with EBSCO today and they're enthusiastic about the idea. Their main question was around request load - do you have any data/estimates about how many daily or monthly requests Copyvios makes?

The other topic we talked about was how pulling the text through would work (or not). EDS has access to all these databases to index for searching, but not necessarily for displaying full text. Even if they did, that would be for subscribing customers so there would be some concern about pulling the full text through to display publicly in the tool. It might be the case that they could return some information about finding a match in a source, but perhaps not display the actual matched text directly. That's something we'll need to get more clarity on with them, but perhaps even if that is the case we could make some UI changes to highlight that a match was found in EDS, and the relevant URL, but not display the matching text? Happy to think that through with you.

If this still sounds feasible to you I'd be happy to copy you into our email thread so you could ask any more specific questions you might have. Samwalton9 (WMF) (talk) 16:25, 5 February 2024 (UTC)[reply]

@Samwalton9 (WMF): Sounds good, thanks for the update! We can definitely indicate a match without including the full text if needed. There is already some support in the tool for this with the Turnitin option.

Regarding request rate, the tool checks about 1,200 articles per day or 36,000 per month. I'd be surprised if that's too much for them, but we could make the new functionality opt-in like Turnitin, so users have to check a box to use EDS which will drastically reduce the rate (the Turnitin feature is used only 100 times/day). — The Earwig (talk) 16:54, 5 February 2024 (UTC)[reply]

@The Earwig Thanks for the data! I remember reading somewhere that the tool makes multiple requests per article check, is that right? I wonder if you have a sense of how many actual API requests are being made? Samwalton9 (WMF) (talk) 13:05, 6 February 2024 (UTC)[reply]

@Samwalton9 (WMF): Yes, that's right – up to 8 per article, depending on page size, but again, configurable. Altogether for Google Search the number is under 10k for most days. — The Earwig (talk) 14:41, 6 February 2024 (UTC)[reply]

Great, thanks! I've cc'd you on an email. Samwalton9 (WMF) (talk) 15:36, 6 February 2024 (UTC)[reply]

Hi! I reached out to Σ by email about lowercase sigmabot III, which had not been archiving anything (with the exceptions of AN and ANI) since last week. They responded (by email) saying Please reach out to Earwig for this issue. The crontab was erased somehow, which means that it's no longer running the bot on its schedule. I'm not sure what changed but I think he will know where to look and that For the time being I just kicked it off manually. Thank you for any insight you might have! HouseBlaster (talk · he/him) 15:07, 28 February 2024 (UTC)[reply]

Thanks for letting me know. I'll take a look at this. — The Earwig (talk) 15:29, 28 February 2024 (UTC)[reply]

It's not clear what the original issue was, but I've jiggled things a bit, so if we're lucky it won't happen again. — The Earwig (talk) 16:29, 28 February 2024 (UTC)[reply]

Thank you! HouseBlaster (talk · he/him) 17:05, 28 February 2024 (UTC)[reply]

News and updates for administrators from the past month (February 2024).

Administrator changes Sdkb The Night Watch East718 Isabelle Belato Mzajac Staecker Stan Shebs Sugarfish Tamzin

Bureaucrat changes SilkTork

Guideline and policy news

• Phase I of the 2024 RfA review is now open for participation. Editors are invited to review, comment on, and propose improvements to the requests for adminship process.
• Following an RfC, the inactivity requirement for the removal of the interface administrator right increased from 6 months to 12 months.

Technical news

• The mobile site history pages now use the same HTML as the desktop history pages. (T353388)

Miscellaneous

• The 2024 appointees for the Ombuds commission are だ＊ぜ, AGK, Ameisenigel, Bennylin, Daniuu, Doǵu, Emufarmers, Faendalimas, MdsShakil, Minorax, Nehaoua, Renvoy and RoySmith as members, with Vermont serving as steward-observer.
• Following the 2024 Steward Elections, the following editors have been appointed as stewards: Ajraddatz, Albertoleoncio, EPIC, JJMC89, Johannnes89, Melos and Yahya.

Hi The Earwig,

I would like to request that talk pages for articles within the scope of WP:WBB be tagged with both

Basketball: Women's NA‑class

This article is within the scope of WikiProject Basketball, a collaborative effort to improve the coverage of Basketball on Wikipedia. If you would like to participate, please visit the project page, where you can join the discussion and see a list of open tasks.BasketballWikipedia:WikiProject BasketballTemplate:WikiProject BasketballBasketball articles NA This article has been rated as NA-class on Wikipedia's content assessment scale. This article is supported by the Women's basketball task force.

Women's sport: Basketball NA‑class

This article is within the scope of WikiProject Women's sport (and women in sports), a WikiProject which aims to improve coverage of women in sports on Wikipedia. For more information, visit the project page, where you can join the project and/or contribute to the discussion.Women's sportWikipedia:WikiProject Women's sportTemplate:WikiProject Women's sportWomen's sport articles NA This article has been rated as NA-class on Wikipedia's content assessment scale. This article is supported by the Women's basketball task force.

Do you need additional information and/or should I post this request somewhere else?

Thank you, Hmlarson (talk) 23:20, 5 March 2024 (UTC)[reply]

Hi Hmlarson, you'll need to define what "within the scope of WP:WBB" means in order to run the bot. — The Earwig (talk) 01:52, 6 March 2024 (UTC)[reply]

Can you do any article tagged with a subcategory of Category:Women's basketball? Hmlarson (talk) 18:58, 11 March 2024 (UTC)[reply]

@Hmlarson: OK. Subcats are sometimes tricky because of unexpected relationships (a subcategory of a subcategory a few levels deep sometimes has little relationship with the original category), but I reviewed this situation, and it looks mostly fine.

I'll have the bot generate a list of pages it would tag, and we can double-check those. It'll take me a few days.

Separately, there is a requirement that you mention on the WikiProject talk page that you want to run this tagging job, in case there are any objections.

Thanks! — The Earwig (talk) 03:43, 12 March 2024 (UTC)[reply]

Thank you. Sounds good. I've posted the notice here. Hmlarson (talk) 17:18, 12 March 2024 (UTC)[reply]

Hi The Earwig - Any chance you have you can provide an ETA on this request? Thank you! Hmlarson (talk) 20:01, 1 April 2024 (UTC)[reply]

Hi, it could be a WP:THURSDAY thing but the revdel-respoder script seems to have a problem today. I keep getting a message "Sorry! revdel-responder failed to parse the page content". I'm not good enough at interpreting the console to work out what's gone wrong. Nthep (talk) 11:44, 14 March 2024 (UTC)[reply]

not sure if anything has happened during the day but, it seems to have resolved itself. Nthep (talk) 18:49, 14 March 2024 (UTC)[reply]

Thanks for letting me know, Nthep. It's possible that was some intermittent error. If you run across it again, let me know the page, or send me the text from the console (right click -> Inspect -> "Console" tab, there should be a line starting with "Error while parsing page content"). — The Earwig (talk) 03:48, 15 March 2024 (UTC)[reply]

News and updates for administrators from the past month (March 2024).

Administrator changes

• Kbdank71
• Kosack
• NrDg
• TLSuda

Guideline and policy news

• An RfC is open to convert all current and future community discretionary sanctions to (community designated) contentious topics procedure.

Technical news

• The Toolforge Grid Engine services have been shut down after the final migration process from Grid Engine to Kubernetes. (T313405)

Arbitration

• An arbitration case has been opened to look into "the intersection of managing conflict of interest editing with the harassment (outing) policy".

Miscellaneous

• Editors are invited to sign up for The Core Contest, an initiative running from April 15 to May 31, which aims to improve vital and other core articles on Wikipedia.