[06:31:51] Welcome bro (re @Feeglgeef: Welcome!) [09:35:01] This is incorrect. As I wrote it should be the episode item QID there instead. : https://tools-static.wmflabs.org/bridgebot/4de40a8f/file_80547.jpg [09:42:35] Should it? The abstract article is calling for Q61220733 (which might not be optimal, but that is what it is doing). (re @Npriskorn: This is incorrect. As I wrote it should be the episode item QID there instead.) [09:43:31] I know, but that's not the one missing a sense, it has no links to any lexemes currently and yet the function works for some languages. (re @Jan_ainali: Should it? The abstract article is calling for Q61220733 (which might not be optimal, but that is what it is doing).) [09:44:46] Yes, see our previous discussion about exactly this here: https://t.me/Wikifunctions/32595 (re @Npriskorn: I know, but that's not the one missing a linked sense, it has no links to any lexemes currently and yet the function works for s...) [09:45:41] In Swedish we need "en" or "ett" which English doesn't so it can fallback to the label and not require a lexeme. (re @Jan_ainali: Yes, see our previous discussion about exactly this here: https://t.me/Wikifunctions/32595) [09:50:59] https://www.wikifunctions.org/view/sv/Z26671?call=%7B%22Z1K1%22%3A%22Z7%22%2C%22Z7K1%22%3A%22Z26671%22%2C%22Z26671K1%22%3A%7B%22Z1K1%22%3A%22Z6091%22%2C%22Z6091K1%22%3A%22Q7093205%22%7D%2C%22Z26671K2%22%3A%7B%22Z1K1%22%3A%22Z6091%22%2C%22Z6091K1%22%3A%22Q61220733%22%7D%7D : https://tools-static.wmflabs.org/bridgebot/ce071af2/file_80548.jpg [09:51:39] https://www.wikifunctions.org/view/sv/Z26671?call=%7B%22Z1K1%22%3A%22Z7%22%2C%22Z7K1%22%3A%22Z26671%22%2C%22Z26671K1%22%3A%7B%22Z1K1%22%3A%22Z6091%22%2C%22Z6091K1%22%3A%22Q7093205%22%7D%2C%22Z26671K2%22%3A%7B%22Z1K1%22%3A%22Z6091%22%2C%22Z6091K1%22%3A%22Q1983062%22%7D%7D [09:51:39] the general "episode" works : https://tools-static.wmflabs.org/bridgebot/36a9a13a/file_80549.jpg [09:53:47] the item has a swedish label : https://tools-static.wmflabs.org/bridgebot/01a95520/file_80550.jpg [09:54:18] this is the english implementation [09:54:20] https://tools-static.wmflabs.org/bridgebot/609c8186/file_80551.jpg [09:56:51] here is the Swedish implementation. I'm thinking we should simplify it to call a function for en/ett instead of using an if. : https://tools-static.wmflabs.org/bridgebot/b5e4fe08/file_80552.jpg [09:56:52] I think it would be better to rewrite the abstract article. Instead of saying "X is a Y", it could say "X is a Y from Z" [10:01:34] There already is an implementation with a en/ett function. But its slightly slower so not the default. Anyway this failure to render is a bug, Q61220733 has a swedish label so it should render, but it doesn't. [10:01:35] The functions are correct and the data in Wikidata is correct AFAICS. [10:05:34] I don't think it is a bug. From the label only, we cannot know if "One Small Step är en Star Trek-avsnitt" eller "One Small Step är ett Star Trek-avsnitt" is what should be rendered. (re @Npriskorn: There already is an implementation with a en/ett function. But its slightly slower so not the default. Anyway this failure to re...) [10:06:20] So how do we improve the situation? (re @Jan_ainali: I don't think it is a bug. From the label only, we cannot know if "One Small Step är en Star Trek-avsnitt" eller "One Small Step...) [10:09:11] No, we write more generic abstract articles. (re @Npriskorn: So how do we improve the situation? Create lexemes for all items? 🤪) [10:11:26] Instead of taking Q61220733 as an argument, it could take Q1092 and Q1983062. [10:15:37] I think this would be a good test case for Z32982. (re @Jan_ainali: Instead of taking Q61220733 as an argument, it could take Q1092 and Q1983062.) [10:27:17] I think this is Davide’s fourth point here [[Wikifunctions:Project chat#How to handle items without lexemes]]. For the general case in many languages, we effectively need to parse the label to determine the head noun (typically). This is not straightforward, but an approximate match between a grammatical noun in the label and a label, alias or lexeme form representation [10:27:17] for a P [10:27:17] 31 item would appear to have a reasonable chance of success, particularly for Germanic languages. (re @Npriskorn: So how do we improve the situation? Create lexemes for all items? 🤪) [12:40:21] This solution cannot be universal, since the question of which items should and shouldn't be used on Abstract Wikipedia can become very tricky. For example, English-speaking contributors could think that Q39201 is a good item to use, (since "pet" is a single word in English) while Italian-speaking contributors could think that it is better to decompose it in [12:40:21] Q729 and Q16514899, s [12:40:22] ince in Italian "animale da compagnia" are 2 different words (plus a preposition). The same thing, with inversed roles, could happen for example for Q2307297 (re @Jan_ainali: No, we write more generic abstract articles.) [12:42:05] Sure, not truly universal, but we could avoid things that are compounds of proper nouns in all languages. (re @dvd_ccc27919: This solution cannot be universal, since the question of which items should and shouldn't be used on Abstract Wikipedia can beco...) [12:44:22] How do we know that it is a compound of proper nouns in all languages? I'd bet that no English-speaking contributor could guess that in Italian (and some other romance languages) Q2307297 is represented by a single word (re @Jan_ainali: Sure, not truly universal, but we could avoid things that are compounds of proper nouns in all languages.) [12:46:01] You don't need to know it for all, it should be enough if it is in one. And I claim that it _is_ clear from the Wikidata item, especially in our example case. (re @dvd_ccc27919: How do we know that it is a compound of proper nouns in all languages? I'd bet that mang English-speaking contributor couldn't g...) [12:46:26] Here comes another problem: what does "from" mean? The episodes did't physically travelled from a place called "Star Trek" to the current position (re @Jan_ainali: I think it would be better to rewrite the abstract article. Instead of saying "X is a Y", it could say "X is a Y from Z") [12:48:14] I agree that this case it is obvious, but what I was evidencing is that this is by no way a universal solution (re @Jan_ainali: You don't need to know it for all, it should be enough if it is in one. And I claim that it is clear from the Wikidata item, esp...) [12:49:09] I’m not sure we’d want to. It seems natural to classify items in terms of their nearest superordinate, as generally available from P31 or P279. The trickier part is exposing the head of the noun phrase, which may mean ascending a P279 hierarchy, with no guarantee of success. (re @dvd_ccc27919: How do we know that it is a compound of proper nouns in all languages? I'd bet [12:49:10] that [12:49:10] mang English-speaking contributor couldn't g...) [12:50:12] I think you just are enhancing my proposal, because it seems like wine harvesting wouldnt include a proper noun in any language. (And if it does, then we can make the sentence more generic again.) (re @dvd_ccc27919: I agree that this case it is obvious, but what I was evidencing is that this is by no way a universal solution) [12:51:03] "From" was just an example, it should obviously be more abstracted depending on what the statement is trying to claim. (re @dvd_ccc27919: Here comes another problem: what does "from" mean? The episodes didn't physically travel from a place called "Star Trek" to the ...) [12:52:32] It was just my frustration on seeing in Abstract Wikipedia functions like Z33975 to express also this kind of sentences (re @Jan_ainali: "From" was just an example, it should obviously be more abstracted depending on what the statement is trying to claim.)