Wikidata talk:Lexicographical data/Archive/2020/11

This page is an archive. Please do not modify it. Use the current page, even to continue an old discussion.

in Netherlands, in the Canada, in United States, in the Belgium .. etc - with or without preposition (English); which preposition (other languages)

At Wikidata:Property_proposal/préposition, we tried to figure out how to add this to Wikidata. The question varies from one language to another. Please provide your input. @Bouzinac, Mahir256, Matěj Suchánek:

Seems I forgot to sign my comment from Oct 28 .. so no pings @Bouzinac, Mahir256, Matěj Suchánek: --- Jura 08:49, 4 November 2020 (UTC)

German adjective

We need to have some agreement on how to model specific kinds of word. For example for German adjectives there are currently several ways to model them:

1. Three comparisons (positive, comparative, superlative) with no inflection data - e.g fremd (L295097)
2. Three comparisons (positive, comparative, superlative) * Four gender/plurality (masculine, feminine, neuter, plural) * (Three declensions (strong, weak, mixed) * Four cases (nominative, genitive, dative, accusative) + predicative) = 156 forms - used in English and German Wiktionaries
3. Three comparisons * (Three gender * Two plurality * Three declensions * Four cases + predicative) = 219 forms - used in the Flexion project

Currently there are no German adjectives with full inflections but many with no forms at all such as fest (L1776).--GZWDer (talk) 08:37, 4 November 2020 (UTC)

Not sure if Wikidata is well equipped for languages that have many (often theoretical) forms.

Depending on your interests, you could start out with the basics like L1776.

If someone is actually interested in the 218 other forms, they could add them once it's well supported by the extension. --- Jura 08:53, 4 November 2020 (UTC)

@GZWDer: What problems and perks has each solution? Could you do temporary example on 2 lexemes to help visualise the situation? Without more details, I see no reason not to add explicitely all forms, at least I see no problem with that (plenty is no plague). Lexemes can easily deal with 219 forms ; our most extreme case is eraman (L222707) with 1274 forms and we have already 471 lexemes with more than 200 forms. A galon, VIGNERON (talk) 14:45, 4 November 2020 (UTC)

Missing language question

You guys might know the answer to this—I'm trying to add the title (P1476) to the newspaper Ojibwe Inaajimowin (Q101823350), that title being in the language Ojibwe (Q33875), which isn't listed in Help:Wikimedia language codes/lists/all. So do I need to start some process to add the language, and/or should I just add the title as undetermined language (Q22282914)? --Struthious Bandersnatch (talk) 18:00, 15 November 2020 (UTC)

Struthious Bandersnatch: I would do both. Using the code 'mis' (and not 'und', it's not undetermined but just without a code, so language without a specific language code (Q22283016)) for now as their is no other solution and meanwhile, ask for a code (and if/when the code is accepted, then change mis to oj) using this newspaper as an example. Cdlt, VIGNERON (talk) 08:47, 17 November 2020 (UTC)

Thanks for replying, VIGNERON! I've been through the various help pages but haven't been able to determine—do you know where I'd want to make the request to add the language code? Could it just be Wikidata:Contact the development team, perhaps? Cheers, Struthious Bandersnatch (talk) 11:18, 17 November 2020 (UTC)

@Struthious Bandersnatch: ah yes, obviously, sorry. Here is the page for explanation and the link for how and where ot ask Help:Monolingual text languages. Cheers, VIGNERON (talk) 11:33, 17 November 2020 (UTC)

Problem with entering translation

The dropdown suggest for entering a Lexeme Sense translation has worked in the past. But cannot get it to work today. --Thadguidry (talk) 17:10, 20 November 2020 (UTC)

Did it? I wish so. But try to add "-S1" to lexeme ID. --Infovarius (talk) 19:49, 21 November 2020 (UTC)

@Infovarius: Ah, you are right, it needs to be a Sense, not a "word". Hmm, the definition on translation (P5972) should match the scope and constraint (which is indeed a Sense), so I'll quickly fix up the definition of translation (P5972) DONE! --Thadguidry (talk) 17:08, 22 November 2020 (UTC)

Some articles about lexicographical data

Hi!

Here are the Proceedings of the XIX Congress of the European Association for Lexicography. This conference was planed for September 2020 but postponed. At least there is some great readings about lexicographical linked data. Noé (talk) 15:53, 24 November 2020 (UTC)

Importing missing pronunciations by bot

Hi, does anyone here know where I can find a bot to run an import on missing pronunciations that exist in Wiktionary or on Q-items but is missing on lexeme forms?--So9q (talk) 10:37, 26 November 2020 (UTC)

Semi-automated import of missing lexemes based on the Q-items with most statements

Hi, we are still missing a lot of useful lexemes in the languages I know (including English).

I would like to have python script that enables me to easily input new lexemes inclusive forms, senses and pronunciation on the command line. If anyone has any code to share (I know about lexdata, but I would need some python code also to fetch Q-items by number of statements also) I would be very happy. :)

The background for this is that I got tired of the cumbersome, inefficient JS-ridden interface of Wikidata and Lexeme Forms and I think I could increase my work efficiency of at least 500% with a well written script (it now about 10 minutes to input 10 lexemes including senses and pronunciation and that could be reduced to a few seconds per lexeme). E.g. for Esperanto words that are extremely regular Lexeme Forms still require me to type everything, which is just a bad user experience IMO. I could create lexemes for all Esperanto labels in Wikidata in a few days with a good tool.--So9q (talk) 10:53, 26 November 2020 (UTC)

@So9q: Have you tried batch-editing with QuickStatements? In addition, the SPARQL query service is useful for finding lexemes that lack senses, statements, etc., which you can then add via QuickStatements. —Ringbang (talk) 23:25, 26 November 2020 (UTC)

Does QS support lexemes properly now? In that case it might be a viable way forward even though I don't like the tool much. Last time I tried it (a year ago or so) it did not work, and I tried contacting Magnus Manske via Twitter and Wikidata and failed. I generally dislike the management of QS, it seems to me that it is unclear which is the main repo (the github or the bitbucket one?) and a lot of the issues has seen no interaction by the author of the tool https://bitbucket.org/magnusmanske/wikidata-todo/issues?q=quick It has a lot of potential as a high quality tool, but in this state of unclear management or interest from the author I'm inclined to find other ways to edit Wikidata efficiently. If anyone here is using the tool, I suggest you step up and find ways to improve it with or without the Magnus in a leading role and document it properly and keep the information page https://www.wikidata.org/wiki/Help:QuickStatements up to date and mark the old one here as deprecated (it contains links to an old version of the tool, for example) https://www.wikidata.org/wiki/User:Magnus_Manske/quick_statements2.--So9q (talk) 10:09, 28 November 2020 (UTC)