Talk:Z26039
Question about the name of this function
What should mean 'article-less' in the context of this function? I think it's a very confusing name, considering that in many languages (for example in Italian) the output of this function should contain the article, without considering languages that don't have any articles. If I understand correctly, even in English "The violin is an instrument." sould be a valid output of this function. Am I wrong? Dv103 (talk) 18:27, 24 July 2025 (UTC)
- Yes, the architecture is unclear. Somehow, it’s the language and the item that determine the appropriate form. For example, “water” requires no article in English but would need a definite article in Italian or French. I think musical instruments are a bit unusual in using the definite article rather than the indefinite, and (in English, at least) we tend to do the same for species but use indefinite singulars or plurals for collections of species. I don’t think this can be directly inferred from Wikidata, currently. But, given that it varies by language (and perhaps also just by editorial style), it’s not clear how it would be represented in abstract content. GrounderUK (talk) 19:27, 27 July 2025 (UTC)
- For now I've interpreted this function as having the entity in a definite form, while subject is kind of (Monolingual text) (Z26095) having the entity in an indefinite form. I've tried to implement this function for Italian, trying a couple of heuristics to guess whether the definite article is definite or not (but the function doesn't work and I don't know why).
- If this is the correct meaning of this function, we should rename it to something like "Definite entity instantiating fragment". In all cases, I think the current name is very misleading. Dv103 (talk) 09:23, 28 July 2025 (UTC)
Allowed entities
I think we should limit the tipologies of allowed entities, for example not allowing people (because usually languages treat people very differently from other things). Dv103 (talk) 21:45, 4 September 2025 (UTC)
- I think you’re right but I’m not sure what categories we might use. Perhaps we want to distinguish between named subjects with or without natural and/or grammatical gender and/or number (not to mention his/her/its/their current state). In other words, whatever tends to drive any inflection of the subject and the copula across a range of languages. This is possibly where Wikifunctions:Project chat#Decision tables for natural language generation is coming from (or heading towards)? @Hogü-456 GrounderUK (talk) 23:12, 4 September 2025 (UTC)
- I have not thinked about this before and so it is good you mentioned it. I think it is better to extend the function and implement the rules to give the correct output when it is about living beings. At least in German sometimes also animals are treated like humans when talking about them. In the decision table I showed how to derive the undefined article ein or eine. It depends on the gender of the second variable part. There is then an additional rule to apply before to derive the relevant form of the second variable from the gender of the first input. I have not added this rule to the decision table. Hogü-456 (talk) 17:57, 5 September 2025 (UTC)
- Another way of dealing with it would be to make a subfunction specific to named subjects / persons. Usually editors should use the most specific appropriate function. But we could also test for the type here and call the subfunction where necessary. Similar issues will surely occur for functions like Z36218 which try to deal with an omnibus of properties, but many properties will need special handling (in that case different languages may split to subfunctions differently). --99of9 (talk) 04:17, 27 July 2026 (UTC)
- I have not thinked about this before and so it is good you mentioned it. I think it is better to extend the function and implement the rules to give the correct output when it is about living beings. At least in German sometimes also animals are treated like humans when talking about them. In the decision table I showed how to derive the undefined article ein or eine. It depends on the gender of the second variable part. There is then an additional rule to apply before to derive the relevant form of the second variable from the gender of the first input. I have not added this rule to the decision table. Hogü-456 (talk) 17:57, 5 September 2025 (UTC)
Applicable to past tense?
Similar to the section above, my question is whether this is supposed to handle former entities? We do not currently have any test cases like this. For example Z26039(Aristotle, philosopher, En) should give "Aristotle was a philosopher." Or should that be handled by a different function? I favour carving it off, with a fallback back to this to simplify for any languages that don't distinguish. I'm happy to go with another plan though, as long as we are clear. --99of9 (talk) 04:21, 27 July 2026 (UTC)