Re: [poe] Wording and RDF semantics: Collection - List - Sequence ...

Some additional facts, @nitmws:


>[…] in terms of RDF 1.1 Semantics, see below, they could be a Collection or a Container.
> * Collection in the RDF 1.1 Semantics** - https://www.w3.org/TR/rdf11-mt/#rdf-collections - is defined as
>
> […] 
> * these "head-tail links" implement in fact an order of the items - and uses the rdf:first and rdf:rest relationships for the links.
> * **A Containerner is in RDF 1.1 Semantics** a kind of enumeration (= Things are only members of the Container but have no relationships among them) - with three specific types
> […]

There is a difference between these two, however. Containers (that was defined in the very first RDF version in the last milleneum whereas Collections came in the current one...) are considered as ill-defined by many. The main issue is that they are not "closed"; in an open world paradigm (of which RDF is part of) it is unknown whether another entity is listed somewhere else for that container, if I follow your analogy with enumerations. The RDF Semantics actually warns:

> “… Since the RDF container vocabulary is so limited, many natural assumptions concerning RDF containers cannot be formally sanctioned by the RDF formal semantics. This should not be taken as meaning that these assumptions are false, but only that RDF does not formally entail that they must be true.”
>
> “… If the container is of an ordered type, then the ordering of items in the container is intended to be indicated by the numerical ordering of the container membership properties, which are assumed to be single-valued. However, these informal conditions are not reflected in any formal RDF entailments.”

Bottom line: _we should to avoid using Containers_. Collections are of course fine.

> […]
> * ... but JSON-LD differentiates between unordered arrays (see above) and "ordered collections" (see section 6.11): to make an array ordered the JSON-LD keyword  @list must be (I didn't find quickly a clear statement by JSON-LD if a @list is RDFied to a rdf:List or something else.)

The RDFization (isn't that a nice word?:-) of a JSON arrays is:

- Without a `@list` it is simply a collection of triples sharing the same object and properties.  I.e., it is _**not**_ an `rdf:Bag`.
- With a `@list`, as you properly say, it is converted into an RDF Collection. (I agree it is not easy to find, it is stated in the last bullet item in https://www.w3.org/TR/2014/REC-json-ld-20140116/#h2_data-model)

> […]
> 
> I see this *state of not clear definitions in the ODRL IM:**
> 1. Key issue: I see not strategy to follow the RDF 1.1 terms - which should be a preference as ODRL IM builds on RDF.
> 2 Currently the RDF expression/serialisation of these "collections" is not defined:
>   * AssetCollection
>   * PartyCollection
>   * ... they are not an ordered list and could be implemented following RDF 1.1 terminology by the RDF Container rdf:Bag.

As said above, we shouldn't do that. 

> 3. "Half-baked" is Constraint Class - rightOperand - "list of ..." as it is unclear: is the RDF semantics of rdf:list applied to this list? By what has  been discussed in #255 it could also be an unordered list, an rdf:Bag. And: does an ordered have to be an rdf:List - or could it be also an rdf:Seq?

Same as above.

> 4. unclear is 2.5.2 Logical Constraint Class as it uses JSON-LD terminology for a part of the definition:
>   * the definition of the only property says "... its value is a list of the existing Constraint instances."
>   * further down for the andSequence is stated, it must use the @list keyword in JSON-LD
>   * open issue: may the "list of existing ..." related to "or", "xone" and "and" be unordered lists?
> 
> RDF experts think about that?

On a very high level, an `AssetCollection` can be a set (in the colloquial sense of the word!) of RDF triples, sharing the subject and the predicate. I would agree with you that the term "Collection" may not be the best term used in conjunction with the RDF terminology; that being said, I am a bit worried about changing the term in that late stage, it may be error-prone.

But I am only an RDF expert, sort of… :-)


-- 
GitHub Notification of comment by iherman
Please view or discuss this issue at https://github.com/w3c/poe/issues/256#issuecomment-329118252 using your GitHub account

Received on Wednesday, 13 September 2017 09:53:35 UTC