Talk:Q18225

questions

@Infovarius‎‎, Snipre: if I understand well, Vitamin A exists into 3 molecular forms ? We can't use has part(s) (P527) for this.

has part(s) (P527) would be useful for orange juice, a mixture of molecules, for example, it could be that

⟨ orange juice ⟩ has part Search ⟨ vitamin A ⟩

but it's not the same semantics here ...

I propose

⟨ Vitamin A ⟩ instance of (P31) ⟨ compound that have several chemical variant ⟩

,

⟨ Vitamin A ⟩ subclass of (P279) ⟨ chemical compound ⟩

(of course), and using the property I proposed for this kind of case (and others)

   Under discussion

Description	MISSING
Data type	MISSING
Example 1	MISSING
Example 2	MISSING
Example 3	MISSING

,

⟨ Vitamin A ⟩ disjoint union of (P2738) ⟨ retinol (Q424976)    ⟩
together with (P1706) ⟨ rhetinoic acid ⟩
together with (P1706) ⟨ ... ⟩

: this would mean "every vitamin A molecule is either retinol (Q424976)    or rhetinoic acid or ..."

But I note that β-carotene (Q306135)    does not appear in frwiki article, is it normal it's in the list here ?gi (

Une molécule de bêta-carotène, par hydrolyse de la liaison 15-15′ sous l'influence d'un caroténoïde mono-oxygénase (ββ-carotène 15,15′ mono-oxygénase), donne deux molécules de vitamine A.

, (a molecule of beta carotene when splitted gives two vitamin A molecules, does that mean vitamin A part of beta caroten however ? author  TomT0m / talk page 20:59, 19 July 2015 (UTC)

@TomT0m: According to en:WP "Vitamin A is a group of unsaturated nutritional organic compounds, that includes retinol, retinal, retinoic acid, and several provitamin A carotenoids, and beta-carotene". So more than 3. The problem here is again what is considered as vitamin A. For some sources, like the one I give from en:WP, the precursors of the vitamin A (meaning the molecules which give vitamin A after a small chemical reaction) are considered as vitamin A. Beta carotene is one of this kind of molecules. So the question is to know which definition of vitamin A is used in the French article.

And no need of a non scientific concept like "compound that have several chemical variant". We have already group of isomeric entities (Q15711994) which is not more scientific but this is a well known term in chemistry.

And your additional property is according to my opinion useless if we can specify in the query that we want all instances linked to the item vitamin A (Q18225) using the classification "instance of"/"subclass of" as link network. We just have to link all chemicals defined as vitamin A to the vitamin A item using "instance of". No need of addtional properties for that. Snipre (talk) 21:51, 19 July 2015 (UTC)

@Snipre: That would make the property instance of (P31) unscientific because not well founded. There is a scientific dispute about Is-a (Q1101080)    in artificial intelligence. It's on when we should use subclass of-like and instance of-like relations. There is also scientific disputes on the maths of set theory and so on, who lead to the importance to build logics and inferences to be we well-founded relation (Q338021)   . So please don't make this a personal question :) Well foundedness can be achieved by type/token relation, which is the basics of many ontologies like chebi, and you won't understand them if you don't have that in mind: this gives a principle to KNOW which objects we classify in the ontology, and if retinoic acid is an instance of a subclass of Vitamin A - it's a subclass.

OK for the definitions, although it hilights the fact we need a way to deal with concurrent definitions and interlanguage links in Wikidata. I proposed a scheme with an item for each definition plus one " metaclass, or disambig item definition of Vitamin A", from which any version of the Vitamin A item definition would be an instance, which would help to find the other definitions and help generate the interlanguage links. I'll add the usecase to WD:XLINK.

OK for the item, I just was looking for him and just did not have the right label. I don't know how it's unscientific :)

Did you read the whole discussion about (disjoint) union of ? It's way more powerful than you seem to imply, and the usecase about nuclide. For example the set of atoms can be partitioned into subset of atoms with the same atomic number, OR by other types or nuclide. This property can express that. Here it have the power to tell Exactly which definition is used and which compound are considered Vitamin A.

For example (Pardon me I'd not be exact, it's just to be an example)

⟨ Vitamin A without precursor ⟩ disjoint union of (P2738) ⟨ retinol ⟩
together with (P1706) ⟨ retinal ⟩
together with (P1706) ⟨ retinoic acid ⟩

⟨ Vitamin A with precursors ⟩ disjoint union of (P2738) ⟨ retinol ⟩
together with (P1706) ⟨ retinal ⟩
together with (P1706) ⟨ retinoic acid ⟩
together with (P1706) ⟨ ...several provitamin A carotenoids ⟩
together with (P1706) ⟨ beta-carotene ⟩

We would add

⟨ Vitamin A without precursor ⟩ instance of (P31) ⟨ Vitamin A definition ⟩

and

⟨ Vitamin A with precursor ⟩ instance of (P31) ⟨ Vitamin A definition ⟩

We just have to find a way to add a statement to Vitamin A definition to mark it as a concurrent definition metaclass, and to code this in the future language link template ... maybe a property to be able to find the definitions with only arbitrary access (maybe union of would do the trick) and one more problem solved :)

Seriously this whole discussion about instance of (P31) and classes is not here to bother you, it just gives solid and it really helps to think on how to organize things here. The scheme I propose is serious and the solution I propose to solve various problem like interlanguage links and definitions variants would be a lot harder without solid foundations. This will help Wikidatians project wide to speak the same language and find consistent solutions to problems ... author  TomT0m / talk page 10:47, 20 July 2015 (UTC)

Before starting a "gas factory" (French people will understand), we can just check if the French definition is validated by any references. Snipre (talk) 14:25, 21 July 2015 (UTC)

@Snipre: Of course. But please note that it's pretty likely that this will be needed in some cases anyway and - for the property proposal - it started long before this usecase showed up.

Also note that the definitions <-> class of definitions structure when concurrent definitions exists in just very similar to the Article with the same name <-> disambig article in Wikipedia, it's not that complex. author  TomT0m / talk page 13:10, 23 July 2015 (UTC)