Talk:DeepL Translator

Latest comment: 1 year ago by Mellohi! in topic Requested move 24 January 2023

Number of translated languages

edit

Why are there two standard versions of Portuguese and English counted as separate languages? I think it is slightly misleading and the actual number of translated languages should be 29 (with an option of selecting a variety in these two languages). (Jan)

Very incomplete article

edit

The article is so superficial that it really looks like a "lose weight fast" scam ad. Some questions that should be addressed are:

  • What group (business? academics? ??) is developping DeepL and where?
  • Which scientific papers describe the underlying technology a bit more specifically?

PhS (talk) —Preceding undated comment added 06:25, 25 January 2020 (UTC)Reply

As of today, I think the article sufficiently explains who is developing it and that the technology is a trade secret, while describing the general strategy. (I'm not an expert in the field.) Fourthark (talk) 01:50, 16 May 2021 (UTC)Reply

Conflicting stats

edit

I was trying to improve the wording in the stats section, because it sounds awkwardly German. That section contains the sentence

> In Germany it ranks 276th, with 30% of the site's users concentrated, followed by France, Switzerland, Spain and Italy.

"Concentrated" is very awkward here but I can't fix it because I don't know if it means something different from the later sentence

> As of 12 May 2019, 21.8% of traffic comes from Germany, 17.4% from France, 10% from Spain, 7.9% from Switzerland and 4.1% from Poland.

If the same meaning is intended, the first sentence should be deleted because it is out of date. Fourthark (talk) 01:56, 16 May 2021 (UTC)Reply

Reception POV

edit

@AngryHarpy: The reception section wording is a remnant from the 2019 article, which was promotional. It quotes positive aspects mentioned by press, while disregarding the articles conclusion. E.g. in the case of Le Monde, concluding that "Google was still a far surperior service", per Connexion translation. More academic sources that thoroughly evaluate the service are needed.

The section also needs restructuring, e.g. the 7 sources used for blind tests contain reviews that can be used throughout the section. Alexa and Similarweb statistics have been deemed un-encyclopedic AFAIK.

Btw the privacy section can probably be removed, original research and primary source. IgelRM (talk) 13:54, 8 July 2021 (UTC)Reply

IgelRM, Alexa's considered acceptable per WP:ALEXA, even though it's largely useless here and maybe shouldn't be, so it's gonna be a resounding eh on my part. I've removed the Privacy section as suggested by you, seems like a reasonable move. Your issues with the Reception section seem also well justified; I only came upon this by coincidence and wanted to prod a little. The criticism from the LeMonde-piece and anything else from around that time should probably be put into perspective for referring to an early version of the service, and the site's usability seems to have unambiguously improved since then, but yeah, some academia'd be nice. AngryHarpytalk 16:40, 8 July 2021 (UTC)Reply
AngryHarpy, this has been the most positive interaction I had on WP, so thanks.
Ok, I have read the Alexa part. Though the additional Similarweb estimate graph still doesn't feel right.
You asked for more examples and SPIEGEL wrote that they had an overall positive impression with good conjugation and sentence structure while there have been some issues with some translations. ("Allerding hakt es auch bei DeepL noch an manchen Stellen, einige Übersetzungen waren sprachlich oder grammatikalisch fehlerhaft. Insgesamt überzeugt der Onlinedienst aber mit der richtigen Konjugation von Verben und stimmigen Satzkonstellationen."). This is all from 2017, but the section notes "reception in 2017" (newer reception would be useful of course). So this section needs the NPOV template in my opinion. IgelRM (talk) 15:59, 16 July 2021 (UTC)Reply

Correction of errors

edit

I am working for DeepL (just to be very transparent regarding COIs). I have first suggested a long edit on December 3, fixing several errors in the articles, but also trying to improve longer paragraphs. Probably this was a mistake. The edit was reverted two weeks later. Today I tried to only correct three short passages that are clearly wrong. I edited only these three parts, and entered very clear information on why the information was wrong before. Unfortunately my edits have been reverted again, without giving any reason based on the content. How can I correct the wrong information, if everything is reverted right away?

Can someone please have a look at the three little changes I made today, and re-implement them? It is the ambition of Wikipedia that there are no false claims in articles. It would be great if someone could check these three edits and the reasons I gave for the edits. It is about wrong information, and nobody wants false claims to be present in the article. DeepL Repr (talk) 13:58, 18 December 2021 (UTC)Reply

Hi, answering here from my talk page. You are acting in good faith and this article has multiple issues with citations as is. However, conflict of interest editors need to use the Request edit system instead. I saw your first edit had reliable sources you could use in the request form. I only reverted because I did not have time to look throughly at your claims. IgelRM (talk) 16:48, 18 December 2021 (UTC)Reply

Frahlinger

edit

I received a message on my talk page from 84.247.216.86 that the recently added part about Frahlinger is false. Given that was based on quoted UEPO.de paragraph below (also mentioned by Gruenderszene.de), it is unclear which part. My intention was to add background about the resignation; however looking back at this, it appears more in speculation WP:Original research territory. Since there appear to be ethical considerations and the topic is now noted on the article's talk page with this comment, I have decided to remove this part from the article.

Dass Frahling sich eher als Wissenschaftler versteht, der lieber in Ruhe seinen Forschungen nachgeht als sich mit der Hektik des Tagesgeschäfts herumzuschlagen, konnte man im vergangenen Jahr bereits ahnen, als er nicht nur UEPO.de bat, sämtliche Fotos von ihm aus früheren Artikeln zu entfernen. IgelRM (talk) 13:09, 3 August 2022 (UTC)Reply

Requested move 24 January 2023

edit
The following is a closed discussion of a requested move. Please do not modify it. Subsequent comments should be made in a new section on the talk page. Editors desiring to contest the closing decision should consider a move review after discussing it on the closer's talk page. No further edits should be made to this discussion.

The result of the move request was: no consensus. Main debate here seems to be the scope of the article (just the translator or DeepL broadly defined?). It does not seem like a page move is conducive to resolving this. (closed by non-admin page mover)Ceso femmuin mbolgaig mbung, mellohi! (投稿) 05:51, 7 February 2023 (UTC)Reply


DeepL TranslatorDeepL – DeepL is emerging as more than just a Translator, with their new Write page. Also, while the main page redirects to the Translator page, the company itself is not called DeepL Translator but just DeepL. Thus the Wikipedia page should be called DeepL with DeepL Translator redirecting to DeepL. Coldbolt (talk) 12:57, 24 January 2023 (UTC)Reply

  • Oppose, DeepL Write is in beta. Is it has been launched, then withdrawn, now launched again in beta. For me it seems like a public test of an uncertain technology. Also, DeepL redirects its own domain deepl.com to the translator, without mentioning write anywhere. So I would suggest to wait if DeepL Write gains some traction. If yes, DeepL will be more than DeepL translator and the article should be moved. But currently I do not see that.
WikiBrendy (talk) 14:44, 27 January 2023 (UTC)Reply
Oppose - I agree with WikiBrendy's rational but for a slightly different reason. You can debate whether the name of a page should depend on the prospect of future pages, but in particular, if the current page name is the best articulate description that can be given or is the easiest way to find the article, then it should be used, regardless of it's effect on the future naming of articles. If a future article comes a long we can change the name.
Regardless, COMMONNAME should not hold in this case because the DeepL refers to the company and DeepL Translator refers to the tool, and these are two classes of instances. Relspas (talk) 18:33, 29 January 2023 (UTC)Reply
The discussion above is closed. Please do not modify it. Subsequent comments should be made on the appropriate discussion page. No further edits should be made to this discussion.