Community Wishlist Survey 2022/Wiktionary/Import translations from wikidata

Import translations from wikidata

  • Problem: Translations could be further enriched automatically or semi-automatically by allowing the import/conversion of wikidata equivalences in translation boxes. For example, for the entry hypocapnia I added the translations from wikidata and converting them to translation box format was a tedious process.
  • Proposed solution: Automate or semi-automate the import and conversion of wikidata translations with the script that would convert wikidata in the format used by wiktionary.
  • Who would benefit: Any users looking for translations.
  • More comments:
  • Phabricator tickets:
  • Proposer: Spiros71 (talk) 11:55, 12 January 2022 (UTC)

Discussion

  • Concepts in Wikidata should not be used directly in Wiktionary. Lexeme Senses are better because of the shared nature. Concept could be expressed by other words not documented in Wikidata. In my experience with French Wiktionary, it's haphazardous to contribute this way. Noé (talk) 13:19, 12 January 2022 (UTC)
    Please refer to the example I mentioned above. Do the wikidata concepts incorporated look haphazardous? Does the end result look haphazardous? Also, I am not talking about direct use, I am talking of automating their import. The editor would need to finetune if needed. Spiros71 (talk) 13:47, 12 January 2022 (UTC)
    Your example is correct, but there is plenty exception such as animal names (Q321376: is it vernacular or scientific name? It depends.) or abstract nouns (Q7242: the label in French is distinct from the name of the page in French Wikipedia). Concepts doesn't include verbs, adjectives, multiwords expressions, affixes, etc etc, a large part of Wiktionaries' content. A suggestion of a list of potential translations is interesting, but I think it's better to display only a short selection of languages known to the user, to capture good information and avoid vague intuition of possible translation in languages not known by the contributor. Noé (talk) 14:03, 12 January 2022 (UTC)
    I do fully understand your concerns and I agree. It cannot work perfectly in all cases, and some rules would need to be made/automated to improve/limit its scope; on the other hand, it can work on a very good level on a huge number of cases (as the one I illustrated above). This is where my focus is. Spiros71 (talk) 08:02, 26 January 2022 (UTC)
    Fully automatic addition seems bound to end in tears. I like the idea of a list of check-boxes for suggested translations garnered from the Wiktionary and Wikipedia links in Wikidata for languages known to the user, perhaps based on babel boxes or a similar declaration by the user. Maybe one could, for cases like “hypocapnia”, allow the user to add languages they are not familiar with, but place these in categories of automated translations to be checked. Such translations could be added at first with a template that marks them as pending confirmation or even conceals them; a tool to automate checking and confirming them would also be nice to have. Another function that might help with semi-automatic addition is displaying part of the linked Wiktionary/Wikipedia entries in the list of suggestions. PJTraill (talk) 19:01, 29 January 2022 (UTC)

Hello there, thanks for your proposal-- I was curious to understand how/if you envision adding translation boxes as related to this wish Automatic adding of translation boxes. Do you envision that the functionality proposed in that wish would benefit this? Are they unrelated in nature due to this focusing on Wikidata linkage? Thanks in advance! NRodriguez (WMF) (talk) 13:49, 17 January 2022 (UTC)

@Spiros71 Pinging so you can see this, thanks so much! NRodriguez (WMF) (talk) 23:11, 25 January 2022 (UTC)
They are related, but different. This procedure is more advanced as it deals with (semi-)automating importing content (and a number of rules would have to be created for that, for example for language name manipulation and compatibility between the two projects), whereas the other one is (semi)-passive, i.e. either automatically provide (blank) translation boxes in all cases, or make it easier for non-technically inclined users to do so (for example, with a button "Click here to add translation [box]"). As it stands now, it is only amenable to people who have some technical knowledge regarding the addition of translation boxes. I hope this is clear; please do not hesitate to ask for any further clarifications. Spiros71 (talk) 07:54, 26 January 2022 (UTC)

its no idea to automize translations. Take an example beugen, german Verb - five meanings and five different english words to describe the verbs impact. If you take easier translation jobs like a noun, its still the accuracy that is todays problem when humans try to find and enter a correct translation. dvs not a word 'referring to something like that' or meaning 'slightly more' than the word in question or 'just one little special case' of the word in question. In my experience its so, that noone ever will control automated entries. If you let them in (if you give an idividual the chance to massimport), they will be there forever. There can be exceptions, but this is my generel view after a lot of practice in the german wiktionary. Pardon my french :) Got a little bit rusty med engelskan mlg Susann Schweden (talk) 18:50, 28 January 2022 (UTC)

“control automated entries” – perhaps you mean “check”? Sorry if it sounds like nit-picking, but this would clarify what you mean! PJTraill (talk) 18:48, 29 January 2022 (UTC)

Voting

  •   Support Vis M (talk) 20:57, 28 January 2022 (UTC)
  •   Support If semi-automatic, based on user abilities with checking tool PJTraill (talk) 21:08, 29 January 2022 (UTC)
  •   Support Libcub (talk) 00:59, 31 January 2022 (UTC)
  •   Support Semi-automatic import with a human in the loop sounds good. Trey314159 (talk) 00:24, 1 February 2022 (UTC)
  •   Support Amirh123 (talk) 12:30, 1 February 2022 (UTC)
  •   Support Sascha (talk) 06:51, 2 February 2022 (UTC)
  •   Support A semi-automatic helper, sure. Urhixidur (talk) 15:58, 3 February 2022 (UTC)
  •   Support - Darwin Ahoy! 00:25, 5 February 2022 (UTC)
  •   Support Wikidata connection by lexemes are rather useful Thingofme (talk) 09:04, 5 February 2022 (UTC)
  •   Support MiltonLibraryAssistant (talk) 11:58, 5 February 2022 (UTC)
  •   Support --Ciao • Bestoernesto 02:41, 6 February 2022 (UTC)
  •   Support —— Eric LiuTalk 10:52, 6 February 2022 (UTC)
  •   Support Ayumu Ozaki (talk) 04:20, 7 February 2022 (UTC)
  •   Support Tom Ja (talk) 18:10, 7 February 2022 (UTC)