Re: RDF lists/arrays and n-ary relations [was Re: OWL and RDF lists]

Hi Anthony,

Le Sat, Oct 01, 2022 at 10:30:27AM +0700, Anthony Moretti a écrit :

> Yes, this makes sense to me too. I've brought it up a few times, it
> actually seems like people have been more receptive to it lately. I've been
> calling them "composite value types", to which I can add your examples:
> 
>   Coordinates
>   Polygons
>   Fractions
>   Addresses
>   ComplexNumbers
>   ChessPositions
> 
> They don't need an IRI, the value's components and its type are enough to
> uniquely identify the value. A comparison operation can be canonically
> defined for each type.
> 
> DateTime is another example, in my opinion DateTime literals are syntactic
> sugar for an extremely common composite value type.

I agree with this and with the name "composite value type".

I am trying to convey the same idea of values that do not need an IRI
and are more than numbers and strings. Datetime is indeed another perfect example.

> Once you have composite value types I think a class of collections—"value
> collections"—falls out of that, so collections where members are values and
> not IRIs.

"One step at a time" ! ;)

-- 
Nicolas Chauvat

logilab.fr - services en informatique scientifique et gestion de connaissances

Received on Saturday, 1 October 2022 12:26:39 UTC