Re: Question on skos:subject domain

Leonard Will wrote:
> In the draft British Standard for thesaurus construction we have used 
> "document" with a very broad meaning, equivalent to "information 
> resource" - it is just more convenient to have one word rather than two..
> 
> That definition is given in <http://www.willpowerinfo.co.uk/glossary.htm>.
> 
> Not all the terms in that glossary have been agreed by the BS working 
> party, but "document" is unchanged from the current draft standard.
Thanks for the comment. The issue for me isn't so much "what kinds of 
things have subjects?", for which "Documents" is a reasonable, general 
answer as long as you take a broad enough view view of what a document 
is. Rather, my question is why foaf:document, and does skos:subject need 
to state that its domain is foaf:document?

The problem I'm seeing is that if I state in my KB:

ianKB:foobar skos:subject someOnt:Whatever .

then my OWL reasoner will entail:

ianKB:foobar rdf:type foaf:Document .

and hence

ianKB:foobar rdf:type <http://xmlns.com/wordnet/1.6/Document> .

(using Dan's wordnet ontology, which is not, as an aside, the actual 
wordnet ontology my KB is using), and a bunch of other entailed 
statements from the FOAF and wordnet vocabularies.  It's all adding a 
fair amount to the search space the reasoner has to explore, so I 
understand the cost ... I just wasn't clear on the benefit.

For the time being, I've decided to coin my own predicate to play the 
role of linking information-bearing resources to skos:Concepts, rather 
than re-use skos:subject.  But it's a shame not to re-use terminology 
that is already specified.

Regards,
Ian

Received on Monday, 21 February 2005 12:30:10 UTC