RE: Top three levels of Dewey Decimal Classification published as linked data

Hi Ryan,

> > Whereas when you want to refer to a specific language or 
> format, you 
> > have to use a specific URI for an information resource, e.g., 
> > http://dewey.info/class/641/about.de or 
> > http://dewey.info/class/641/2003/about.de. So language _is_ 
> recognized 
> > as part of the domain, but only as part of representations, not of 
> > concepts.
> 
> I don't understand this part. I understand why you ought to 
> have language-specific (and versioned) concepts for the DDC. 
> But then it seems to me that you should have 
> language-specific (and versioned) URIs for the *concepts*, 
> not just for the information resources that represent them. 
> As it stands (and as Ed pointed out), you seem to be 
> asserting that your information resources are SKOS Concepts.

I think this a general difficulty with the SKOS model. What is a SKOS
concept? Is it a thing ("unit of thought") or an information resource?
It seems that it can be both, depending what information you provide and
how you refer to it. For example, it could be argued that the concept
http://id.loc.gov/authorities/sh85137241#concept really refers to the
idea of trees in general, or (probably more in line with common opinion)
to the idea of trees as framed as an element of the controlled
vocabulary LCSH. This thing has (following the definition given in
webarch) as a non-information resource no representation that can be
transmitted over the web. So this concept might have broader and
narrower terms, but can it as a thing have a modification date? Can it
have a history note? (A car can have a color, but only the description
of that car can have an dct:issued date or a dct:language.)

Tennis and Sutton [1] have suggested (for somewhat different reasons) to
split skos:Concept into skos:Concept and skos:ConceptInstance to clarify
what kind of skos:Concept we are talking about; a "thing" or its
instantiation as an information resource.

I ran into the same problems with modeling DDC concepts. Is it possible
for an abstract concept to have a language (e.g.
http://dewey.info/class/641/en/), or, inversely, for a description of a
concept to _not_ have a language?

My tentative answer was to split the DDC universe in abstract things
(http://dewey.info/class/641/) and descriptions of these things
(http://dewey.info/class/641/about). I thought about language as being a
dimension of the description, not the concept, the same way dct:issued
(usually) is about an object's description, not the object itself. By
this definition, a skos:Concept (that produces a 303) can never be
language- (or format-)specific. It can only be time-specific and is
_generic_ in terms of language and format. If the concept has a
language, it has to respond with "200".

Do you think I am on the right path with this conceptualization? Some of
this seems to be part of TAG Issue-63 [2]. The question it comes down to
for SKOS implementors is probably: Given a concept that is a
non-information resource and a concept that is an information resource
describing (providing metadata about) that concept, can both be
skos:Concepts? To come full circle: was it OK to model
http://dewey.info/class/641/ as a skos:Concept in the first place?

Michael

[1] http://www3.interscience.wiley.com/journal/116332034/abstract
[2] http://www.w3.org/2001/tag/group/track/issues/63

Received on Friday, 21 August 2009 21:05:48 UTC