Wikipedia talk:WikiProject Guild of Copy Editors

Probable error

In the "How can you help" section, it shows that there are 0 articles in the backlog. It seems to be an error. Could someone please confirm? TNM101 (chat) 16:51, 10 December 2024 (UTC)

Fixed, TNM101 (I think); the category was renamed. Thanks for the poke and all the best, Miniapolis 21:30, 10 December 2024 (UTC)

Experiment with automatic correction suggestions

I've started an experiment to automatically detect spelling and grammar errors using a script, and I was wondering if some copy editors here might be interested in helping me test it. For this purpose, I've created the page User:Phlsph7/Correction suggestions experiment, which lists correction suggestions for a random selection of newly created articles and articles found at Category:All_Wikipedia_articles_needing_copy_edit. Each suggestion has the following format:

−

Banu Nahd has been an Important tribe In Wadi Hadhramaut migrating to the region from their original homes in Najran and its environs during 1195 AD.

+

Banu Nahd has been an important tribe in Wadi Hadhramaut migrating to the region from their original homes in Najran and its environs during 1195 AD.

Explanation: The words "Important" and "In" should not be capitalized as they are not proper nouns or the beginning of a sentence.

The idea is for a copy editor to review one suggestion at a time and either fix the error in the article if the suggestion is valid or dismiss it if it is not. Afterwards, they use strikethrough on the suggestion with <s>...</s> (e.g. [1]) so other editors know that it has been checked.

I would be interested to hear whether creating this type of list of correction suggestions for others to review and fix could help copy editors improve articles. Perhaps someone could work through a few entries of the list and share their impressions. I'm also open to other ideas on how this approach could be used.

The suggestions were created using an AI model and can include inaccuracies, so copy editors should use their own judgment and not implement them blindly. The script may overlook errors in the article. It focuses primarily on objective errors rather than style and formulation improvements required for polished prose. Additionally, the script only reviews regular paragraphs and ignores mistakes in lists, image captions, and similar elements. Phlsph7 (talk) 17:56, 11 December 2024 (UTC)

Thanks, Phlsph7. I'm pinging Beland, who may want to use this to compile their database reports, and will check out your script ASAP. All the best, Miniapolis 23:55, 11 December 2024 (UTC)

Neat! For the moss project I'm generating lists of possible spelling and punctuation errors like these: Wikipedia:Typo Team/moss/D. Though for a while I think we were fixing typos faster than they were being added, volunteer motivation has dropped off, and now we are going very slowly. I think the problem is that our on-wiki system is just too cumbersome. We rely on the humans to load the article, find the typo, fix it manually, and then also manually update the on-wiki list of typos as to whether or not it's been fixed. My plan was to build a JavaScript page in the style of JWB, which is derived from Wikipedia:AutoWikiBrowser. I'm not currently generating spelling fix suggestions, but I can use an open-source spell checker to do so, and some custom code to suggest punctuation fixes. The idea is that volunteers would load up 100 or 1000 articles to be fixed (or pick at a shared list of a million possible typos) and be able to just click buttons like "Accept suggestion" and "Not English" and "Correct as written" and whatnot. Though they'd probably need to manually edit some instances where the suggestion is wrong, hopefully most of the typos could be handled with a single click, and the system would automatically update the shared todo list somewhere. I'm not sure if we'd be able to continue using wikitext as our backing store; a proper database hosted on Wikipedia:Wikimedia Cloud Services would be a more professional solution. I expect this sort of system would enable volunteers to go maybe 10x or 25x faster, which would be more fun and a lot easier for a smaller number of volunteers to keep up.

Presumably a similar setup would make it quick to act on the suggestions that your script makes, though I'm not sure how interested you'd be in building a more complicated system. How do you actually run your script? I peeked at it briefly but haven't figured it out completely. Is it meant to be installed as a user script and used on-site while logged in to Wikipedia? (If so I'm not sure what activates it.)

I'm not sure what fraction of typos your scripts is catching, but it's possible that if we were to run your script on all the articles on English Wikipedia, it would find tens or hundreds of thousands of problems. Given that it runs through a third-party API, it seems unlikely that we could do that for every database dump (we get two a month), so there's some question as to how we'd decide which articles should be fed in. Using the list of articles that need copyedit is a good idea. Though I think those articles still need to be read end-to-end by a human, automatically fixing some of the typos might make the process faster and possibly more thorough. New articles are another interesting target, though I expect these to be unstable, and much more likely to have typos fixed by random editors compared to the long-lived typos we are fixing by going through all articles systematically.

Many of the errors caught by your system would not be caught by moss (I haven't coded in grammar checking yet, but I have ambitions), but on the other hand, pages with spelling and punctuation errors detected by moss might be the ones worth running through OpenAI. (It would also be interesting to see how many of the spelling errors OpenAI finds.) We could also just slowly run it over the entire list of articles if that would be useful. I can share article lists if that would be helpful. I'm curious what daily rate the OpenAI API would limit us to? -- Beland (talk) 03:49, 12 December 2024 (UTC)

Hi Beland, I wasn't aware of the moss project but it sounds like an exciting approach. I haven't read the documentation, but if I understand the list at Wikipedia:Typo_Team/moss/D correctly, it lists unknown (and probably non-existing) words in articles. For example, the line "1 - Dana Snyman - wikt:schetches" says that the article Dana Snyman has the unknown word "schetches". But the list does not give correction suggestions. Running the SpellGrammarSuggestions script on this article I get

−

Both are collected short-short stories and schetches.

+

Both are collected short-short stories and sketches.

Explanation: The word "schetches" is a misspelling. The correct spelling is "sketches," which refers to short, informal pieces of writing or drawings.

So the script (or a similar script) could be used to provide editors with correction suggestions. Having a JWB-like script to allow reviewers to accept or reject suggestions with one click could significantly speed up productivity for fixing simple typos.

Regarding the SpellGrammarSuggestions script, I added a short info page at User:Phlsph7/SpellGrammarSuggestions explaining how it works. The short version is: you install it as a user script, navigate to an article, click on the link "SpellGrammarSuggestions" in the toolbox under "Tools", and click on "Start" in the control view that just opened. If you use it the first time, you need to add an OpenAI API key by clicking on the button "Add/Remove API key" before pressing the button "Start".

In its current form, the script works on one article at a time, so getting suggestions for long lists of articles would be very cumbersome. However, it shouldn't be too difficult to write another script that receives a list of articles and produces suggestions for each article in the list. Instead of finding new errors, one could also limit the SpellGrammarSuggestions script to only provide suggestions for unknown words detected by the moss script.

There are certain rate limits listed at [2]. I haven't had any problems with them but they might become an issue if we wanted to run the script on a huge number of articles. In that case, cost could also become an issue. The cost of suggestions for one article varies depending on the size of the article and the number of errors found. From my experiments, it is usually a fraction of a cent (US) per article. That might come down even further if we only focus on suggestions for unknown words. Phlsph7 (talk) 11:34, 12 December 2024 (UTC)

Attempted to make SVG icon

Hello, I attempted to make an SVG version of the guild logo using avaliable SVGs File:Writing Magnifying.svg. I had to try and recreate the quill so it's not quite right, as I'm pretty new with Inkscape. Feel free to edit it or make your own. Cmrc23 ʕ•ᴥ•ʔ 09:25, 22 December 2024 (UTC)

Thanks very much; it's an improvement over File:Writing Magnifying.PNG. All the best, Miniapolis 14:41, 22 December 2024 (UTC)

Cheers! I also made a bronze, silver and gold version for all the guild's awarding needs (and a topicon) Cmrc23 ʕ•ᴥ•ʔ 15:00, 22 December 2024 (UTC)

square feet: singular or plural?

@Epicgenius

• there are 169,000 square feet of floor space
• there is 169,000 square feet of floor space

One argument is that there's more than one square foot, so it's plural. The other is that it's a measurement, not a count, so it's singular. Which is right? RoySmith (talk) 19:05, 31 December 2024 (UTC)

Hi @RoySmith:, there are multiple square feet so use "there are ..." – you wouldn't write "there was 27 rows of cabbages" or "there was 27 acres of forest". Depending of the context, you can rephrase it as "The floor space is 169,000 square feet" and neatly stop the argument. Cheers, Baffle☿gab 22:04, 31 December 2024 (UTC)

I recommend using the Convert template which automatically selects the term and provides the metric version. Rublamb (talk) 19:49, 1 January 2025 (UTC)

Is "medicine-articles" clear, comprehensible, and accurate?

The first paragraph of Wikipedia:Identifying reliable sources (medicine) contains this sentence:

Sourcing for all other types of content – including non-medical information in medicine-articles – is covered by the general guideline on identifying reliable sources.

If I were the god of Wikipedia I would change "medicine-articles" to "medical articles" without any discussion, but since I have yet to receive deity status, I will first discuss my concerns with other copy editors, and then, if appropriate, discuss on the article's talk page.

My rationale for changing medicine-articles to medical articles:

1. Comprehensible - I stumbled over medicine-articles when I first read the sentence. (Do they mean medical articles? Or does medicine-articles have a special meaning in this context?) Medical articles is reader-friendly; medicine-articles is not.

2. Clear - The sentence sacrifices parallel structure, which often improves clarity, by using an adjective + noun combination, followed by a noun adjunct-noun word.

Garner notes that "...the reader's expectations are thwarted when a noun is used adjectivally in place of the more usual adjective."

Bryan A. Garner, Garner's Modern English Usage (4th ed. 2016) at 416–417.

3. Accurate - I'm not sure about this one, but medicine-articles does not seem grammatically (or in any other way) correct. I could see medicine-articles as a phrasal adjective if it preceded a noun, but it does not. I called it a noun-adjunct-noun word above, but perhaps it is best described as a noun-noun neologism.

Having said all that, I keep thinking: "I must be missing something since this sentence has existed for quite a while in a major Wikipedia content guideline." What am I missing?

Note: I searched the article's Talk page archives for any discussion about medicine-articles, but did not find anything. -- Mark D Worthen PsyD (talk) [he/him] 17:31, 2 January 2025 (UTC)

Hi @Markworthen:, I would say in this context the word "medicine" is adjectival and the word "articles" is a noun, so the hyphen should be removed. For clarity, you could change "medicine-articles" to "articles about medicine". Just my opinion, though. Cheers, Baffle☿gab 18:40, 2 January 2025 (UTC)

Thanks!

I proposed changing medicine-articles to medical articles on the article's talk page. I will wait a few days (or more) before I edit that sentence (or decide to leave it as is). I referenced this discussion. I realize that discussing something on two talk pages can create problems. Perhaps more technical discussions about grammar, punctuation, and usage could take place here. But I defer to more experienced editors on this point, e.g., maybe this should be discussed only at the WP:MEDRS talk page. Mark D Worthen PsyD (talk) [he/him] 03:45, 3 January 2025 (UTC)