[05:27:53] I see these functions are starting to use QID arguments requiring a "semantic role" and a (semantic) "relation", and that the examples call examples which are instances of Q134506144, mostly documented in WD by @mahir256 . This looks like a key moment of a step toward convergence toward the complexity allowed by his NLG system. Before we use it too much, do others [05:27:53] want to check i [05:27:54] f these functions are using the definitions right, and are well-defined? I was tempted to wonder if the allowed arguments should be lightweight enums, but it looks like there are thousands of options `https://w.wiki/Wh67` . Another more technical issue: the "exact match" links to w3id.org (https [05:27:54] //w3id.org/framester/data/propbank-3.4.0/LocalRole/embody-01.entity-embodied) are cu [05:27:55] rrently broken, so it's hard to see what each "proto-patient" means apart from its label. What other properties should be used to document these, and is there any chance that these links will come back online? (re @u99of9: Z44182 feels incredibly powerful, but so far I've only understood it well enough to blatantly copy the sentence about Japan.) [09:08:23] Tough one! I suppose it’s fine to use unconstrained QIDs to express a relation, but we should expect lexicographic links for them. If that’s not appropriate for the Wikidata community, maybe there should be functions to do the job? [09:08:24] I also have reservations about a list of “adverbs” for relation qualifiers, because that seems to make assumptions about a language’s grammar. Even in English, “administratively divided” would perhaps be more conventionally expressed as “for administrative purposes” and, perhaps better, “has n administrative divisions called [prefecture]s”. Neither enwiki nor Si message> [09:08:25] mple qualifies the division adverbially: [09:08:27] “Japan (https://en.wikipedia.org/wiki/Japan) is divided into 47 prefectures (https://en.wikipedia.org/wiki/Prefecture) (都道府県, _todōfuken_, [todoːɸɯ̥ꜜkeɴ (https://en.wikipedia.org/wiki/Help:IPA/Japanese)] ⓘ (https://en.wikipedia.org/wiki/File:TomJ-Prefectures.ogg)), which rank immediately below the national government and form the country's first level of juris message> [09:08:28] diction (https://en.wikipedia.org/wiki/Jurisdiction) and administrative division (https://en.wikipedia.org/wiki/Administrative_divisions_of_Japan).” [[:en:Prefectures of Japan]] [09:08:30] « Les *préfectures* ou *départements*[1]* du Japon* (都道府県, todōfuken (https://fr.wikipedia.org/wiki/Aide:Japonais)*?*) sont les quarante-sept circonscriptions administratives (https://fr.wikipedia.org/wiki/Circonscription) et territoriales du Japon créées (https://fr.wikipedia.org/wiki/Japon) en 1871 (https://fr.wikipedia.org/wiki/1871). » [09:08:31] [[🇫🇷Préfectures du Japon]] [09:08:33] « Le Japon est subdivisé en quarante-sept préfectures (https://fr.wikipedia.org/wiki/Pr%C3%A9fectures_du_Japon) (ou départements), dont… » [09:08:34] [[🇫🇷Japon [09:08:36] #Subdivisions administratives]] [09:08:37] Perhaps the real challenge here is to comply with the norms of local Wikipedia editions in specific contexts, whilst retaining the flexibility of more general purpose constructions for particular special cases (like the UK, of course). (re @u99of9: I see these functions are starting to use QID arguments requiring a "semantic role" and a (semantic) "relation", and that the [09:08:37] ex...) [09:28:00] Which UK issue are you referring to? [09:34:49] “No one” understands the administrative divisions, unless you stick to pre-1974 counties, which already had a few oddities like the City and County of Bristol (home of Gloucestershire county cricket). 🤷‍♂️ (re @u99of9: Which UK issue are you referring to?) [09:39:39] yeah, France administration is weird (like cantons that are bigger or small / above/under the communes) but from what I've seen, UK is a whole other level (re @Al: “No one” understands the administrative divisions, unless you stick to pre-1974 counties, which already had a few oddities like ...) [09:47:05] You guys can write a function of your own for that! [10:00:28] Maybe start with the islands?? (re @Al: “No one” understands the administrative divisions, unless you stick to pre-1974 counties, which already had a few oddities like ...) [10:13:45] That might also be a problem, but an orthogonal one [[en:British Islands]]. 😎 (re @u99of9: Maybe start with the islands??) [12:57:17] Hi all, a quick question: how can we surface more this page http://abstract.wikipedia.org/wiki/Help:Abstract_Data_guide relative to our tool Abstract Data? where we can feature it in your opinion? [13:01:25] Is it ready to feature already? I am looking at it and the top missing labels and missing lexemes aren't "missing" at all, they should not be created. (re @Sannita: Hi all, a quick question: how can we surface more this page http://abstract.wikipedia.org/wiki/Help:Abstract_Data_guide relative...) [13:02:55] can you explain more about this? (re @Jan_ainali: Is it ready to feature already? I am looking at it for Swedish and the top missing labels and missing lexemes aren't "missing" ...) [13:03:36] These should not get Swedish labels, they should get mul labels. : https://tools-static.wmflabs.org/bridgebot/d1c04c2f/file_84341.jpg [13:03:43] I see [13:04:29] I don't think these should have lexemes in Swedish. : https://tools-static.wmflabs.org/bridgebot/c0aec0b2/file_84342.jpg [13:05:01] I put some similar comments on David's page a couple of weeks ago: https://www.wikifunctions.org/wiki/User_talk:DSantamaria-WMF#Abstract-data_again [13:05:07] Thank you [13:28:09] One heuristic to see if a lexeme should be expected would be to see if *any* language has a lexeme. [13:32:05] Which happens to be what I was thinking for “lexicographic links” for QIDs expressing a relation (“proto-patient” or otherwise) 😏 (re @u99of9: One heuristic to see if a lexeme should be expected would be to see if *any* language has a lexeme.) [13:36:50] On the adverb issue: I think language implementations should feel free to express the sentence in any alternate way that has the same meaning. So even if they are not used as actual adverbs, perhaps the data structure is sufficient to make that meaning clear in general?? (Hopeful...) (re @Al: Tough one! I suppose it’s fine to use unconstrained QIDs to express a [13:36:50] relation, but we [13:36:51] should expect lexicographic links for the...) [13:44:55] I know what you mean, but it’s like “tense” or “sentence”: I don’t think it is ultimately helpful to use linguistic terms for expressing epistemic or semantic relations. (re @u99of9: On the adverb issue: I think language implementations should feel free to express the sentence in any alternate way that has the...) [13:51:14] I have no issue relabeling if that resolves the issue (and if it remains comprehensible for the editors). But I'm more concerned with checking/ensuring the structure is valid in the first place. [13:53:37] In the absence of lexicographic links, I don’t see how it can be. (re @u99of9: I have no issue relabeling if that resolves the issue (and if it remains comprehensible for the editors). But I'm more concerned...) [13:54:54] Which links are missing? I haven't looked deeply. [14:11:44] Ah, I see we have P9970 links for L726 and a Japanese translation added yesterday… I may have to defer to the Wikidata community on that one 🤔 (re @u99of9: Which links are missing? I haven't looked deeply.) [14:39:40] In any event, Q140429971 appears to be the wrong kind of “division” for the equivalent of comprise L14540 (Q1310239 or Q36809769, perhaps). (re @u99of9: Which links are missing? I haven't looked deeply.) [18:03:01] I consider to have some kind of mini-meeting (with presentation and open discussion) on my dictionary project (the project that got me into this channel :) this friday. if anyone is interested, feel free to ping me. [22:44:46] Can you provide a brief abstract of your dictionary project? (re @wmtelegram_bot: I consider to have some kind of mini-meeting (with presentation and open discussion) on my dictionary project (the p...)