On Fri, Feb 01, 2013 at 12:32:41AM +0100, Bernard Vatant wrote:
> "Language" is defined as "A language of the resource", so some of the
> > examples seem to fit better than others:
> >
> > -- the official languages used during an event etc.
> > The resource is an Event, and X is a language of that resource, so it seems
> > to fit. The event has an official language.
>
> The same property "official language" could apply to Country, Organization,
> Project or Event, at first sight.
> But the more I think about it, the more I doubt this property fits really
> under dcterms:language.
> The way dcterms:language is applied to library assets is more the language
> in which the content of the resource is expressed.
> The language of a Country is not a property of the content of the Country,
> it's the language spoken by its citizens, or used by its official documents
> etc. So the valid argument you have below for languages used by a Person
> aplly here also it seems.
A historical footnote: From 1998 to 2006, Language was defined as [1]:
A language of the intellectual content of the resource.
In 2006, the definition was changed to its current definition:
A language of the resource.
with the following justification [2]:
|| 3.1. Replacement of the phrase "the content of the
|| resource" (or "intellectual content of the resource")
|| with "the resource".
||
|| Implementation experience has shown the semantic
|| distinction between a resource and the "content of"
|| a resource to be a source of confusion.
||
|| According to the DCMI Abstract Model, a DCMI metadata
|| description "describes one, and only one, resource"
|| [DCAM]. In the DCMI context, this principle dates
|| back to 1997, when it was known as the "one-to-one
|| principle". When the one-to-one principle is
|| correctly followed, the semantic restriction to
|| "the content of" a resource is in effect redundant.
||
|| Moreover, Dublin Core elements are used to describe
|| things other than document-like resources -- things such
|| as physical objects and abstract concepts. For example,
|| a stuffed animal has a Creator (perhaps a taxidermist,
|| or an artist) but that person is not specifically the
|| creator of "the content of" the stuffed animal. In light
|| of current usage, the meaning of the original distinction
|| between a resource and its content is no longer clear.
||
|| In the opinion of the Usage Board, dropping the words
|| "the content of" from definitions affirms the semantics of
|| these terms as they have been understood in practice and
|| with no significant practical impact on implementation.
||
|| This change is relevant to the definitions of Contributor,
|| Creator, Coverage, Description, Subject, Type, and
|| Language.
[1] http://dublincore.org/documents/2006/08/28/dcmi-terms/#language
[2] http://dublincore.org/usage/decisions/2006/2006-03.dcmes-changes.shtml
> See above. "the resource has a language" seems to me a shortcut to "the
> content of the resource is expressed in some language"
And as I see it, dropping the words "of the intellectual content" from the
definition does not contradict that interpretation.
Taking one more step back, one could perhaps say that in a Semantic Web
context, the words used in an official definition are just part of what really
defines a property. Equally important, and increasingly so over time, is the
sum of contexts in which a property is used. Being able to pull together that
sort of information in an automated way would be extremely valuable. Right
now, for example, it would be great to be able to pull out examples of how
dcterms:language is actually being used.
Tom
--
Tom Baker <[log in to unmask]>
|