W3C home > Mailing lists > Public > semantic-web@w3.org > July 2007

Re: vCard confusion and RDF insufficiency

From: Garret Wilson <garret@globalmentor.com>
Date: Thu, 26 Jul 2007 10:04:46 -0700
Message-ID: <46A8D42E.5030106@globalmentor.com>
To: Harry Halpin <hhalpin@ibiblio.org>
CC: Semantic Web <semantic-web@w3.org>

Harry Halpin wrote:
> I have yet to see anything resembling a substantial objection to using
> rdf:Seq rather than rdf:List. Does *anyone* object to rdf:Seq, which
> seems to be able to take literals as objects with an ordering constraint?


* Bruce: "Everyone I have talked to has discouraged use of rdf:Seq." 
(semantic-web@w3.org 2007-07-26)

* Benjamin: "Yes, that's a general suggestion, which usually coves both 
collections and containers, as they introduce intermediate nodes." 
(semantic-web@w3.org 2007-07-26)

* Dublin Core: "The RDF Container constructs rdf:Bag, rdf:Alt and 
rdf:Seq are no longer provided as an alternative for constructing 
ordered and unordered sets." ( 
http://dublincore.org/documents/dc-rdf-notes/#sect-5 )

* For rdf:Seq, you can have multiple rdf:_3 properties, for example.

* For rdf:Seq, you could have a rdf:_2809 property with no other 
properties, for example.

* There is no way to specify the end of an rdf:Seq list.

> The only objection would be the lack of closure of containers and lack
> of formal semantics or ordering, but given that you cannot reasonably
> put literals in rdf:List, I see no option other than use rdf:Seq.

As pointed out by Sandro, you can indeed have literals in a rdf:List 
(see http://www.w3.org/TR/rdf-schema/#ch_first ). It's just that it's a 
pain to place literals in a rdf:List using RDF+XML serialization.

It seems fundamentally wrong to me to allow a particular serialization 
format of a general model to dictate the construction of an ontology.

>  2) Leave the range constraint out and so allow users have "unordered"
> additional names (and the same question for honorable prefixes and
> suffixes, which I imagine may have ordering constraints in some
> languages, and some order for them seems to be implied by the vCard
> RFC), but remind the users in the spec that for this property if
> ordering is needed, one *should* use rdf:Seq.
> I am leaning towards 2), since I am in general always leaning on letting
> users use complex things (such as rdf:Seq) as optional.

And things come full circle. Remember, before The Compromise, that was 
my proposal for all name elements! Let v:givenName take a literal or a 
list, as the need allows, but there was a lot of opposition to this 
"value-switching". The same with v:familyName and v:additionalName. In 
fact, I've had it implemented this way for months---I was trying to cave 
to get consensus.

Received on Thursday, 26 July 2007 17:04:57 UTC

This archive was generated by hypermail 2.3.1 : Tuesday, 1 March 2016 07:41:58 UTC