W3C home > Mailing lists > Public > public-gld-comments@w3.org > April 2013

Re: dataset and Dataset

From: Richard Cyganiak <richard@cyganiak.de>
Date: Sat, 6 Apr 2013 17:53:45 +0100
Cc: Makx Dekkers <makx@makxdekkers.com>, 'Bill Roberts' <bill@swirrl.com>, public-gld-comments@w3.org
Message-Id: <E3734AF2-48F2-45CD-9F5E-66F1994989F5@cyganiak.de>
To: Phil Archer <phil@philarcher.org>
On 5 Apr 2013, at 12:37, Phil Archer <phil@philarcher.org> wrote:
> I was told by non-Sem Web data modellers that the more general convention is that properties (data type properties) should be nouns and relationships (object type properties) should be verbs.

That's interesting. The only data modelling school that I'm aware of that can be said to more or less follow this convention is a Sem Web school -- Description Logics.

Among UML modellers, OO modellers, DB modellers and XML Schema modellers, nouns as relationships seem to be completely normal, and in fact more common than verbs.

So I'm not prepared to accept the claim made by your non-Sem Web data modellers without some evidence.

> And actually, isn't dcterms:hasPart sufficient for what dcat:dataset does anyway?

So is the proposal is to fold both dcat:dataset and dcat:record into one property dcterms:hasPart? I'd rather keep two separate properties for linking the catalog to its datasets and to its catalog records. It makes queries easier.

Also, conceptually, CatalogRecords certainly are part of the catalog, but I feel that DCAT shouldn't take a stance on the question whether the datasets themselves are “part of” the catalog. Use cases are common where the catalog only *describes* the datasets, but can't really be said to *include* them.

> The definition is "A related resource that is included either physically or logically in the described resource" which sounds rather like a dataset in a catalogue to me.

You have a specific conceptualization in mind here that isn't universal.

Best,
Richard


> 
> Phil.
> 
> On 05/04/2013 11:52, Makx Dekkers wrote:
>> 
>> 
>> I'd like to point to a decision taken by the Dublin Core Metadata
>> Initiative and documented at
>> http://dublincore.org/documents/naming-policy/ :
>> 
>> 
>> 
>> "No DCMI Term Names will be assigned that differ from other Names only
>> in regard to case."
>> 
>> 
>> 
>> Can you refer to cases where vocabularies use the same name for a class
>> and a property with only difference in case?
>> 
>> 
>> 
>> I looked at other vocabularies (FOAF, VOID, SKOS,  GoodRelations, DOAP,
>> MO, RDFS) and none of them seem to do this. I found some examples where
>> they use class name Something and then have a property hasSomething
>> (e.g. GoodRelations).
>> 
>> 
>> 
>> Data Cube is the only other example that I could find where the same
>> names are used with upper and lower case, but I haven't done a deep
>> investigation.
>> 
>> 
>> 
>> I have no opinion either way but it may be good to listen to what
>> implementers say.
>> 
>> 
>> 
>> Makx.
>> 
>> 
>> 
>> 
>> 
>> 
>> 
>> From: Bill Roberts [mailto:bill@swirrl.com]
>> Sent: Friday, April 05, 2013 12:13 PM
>> To: public-gld-comments@w3.org
>> Subject: Re: dataset and Dataset
>> 
>> 
>> 
>> I understand but disagree with Alasdair's point on potential confusion
>> between dataset and Dataset.  The convention of lower case for predicate
>> and upper case for class (usually the range of the corresponding
>> predicate) is now quite well established and is quite easy for new
>> people to learn - because it is widely used.  So sticking to this
>> approach seems fine to me.
>> 
>> 
>> 
>> Bill
>> 
>> 
>> 
>> 
>> 
>> B
>> 
>> 
>> 
>> 
>> 
>> 
> 
> -- 
> 
> Phil Archer
> http://philarcher.org/
> +44 (0)7887 767755
> @philarcher1
> 
Received on Saturday, 6 April 2013 16:54:09 UTC

This archive was generated by hypermail 2.3.1 : Wednesday, 10 October 2018 10:43:22 UTC