Re: Using "Punning" to Answer httpRange-14

On Mon, May 14, 2012 at 1:22 PM, Michael Brunnbauer <brunni@netestate.de> wrote:
> You just conceded that with dcterms:subject there are 4 valid options
> and not a single one. So you think the dcterms vocabulary is "broken" because
> it does not include the three variants of dcterms:subject that somehow relate
> to the content of a URL ?

In http://dublincore.org/documents/dces/ 'subject' is defined as 'The
topic of the resource.' To me that means that if:

  <h2 property="subject">World War II</h2>

appears somewhere in a document, then either the section following
that header, or the entire document, is about world war II. I would
say it exclude using Book subject: <span property="subject">world war
II</span> in a web page about a book. So that reduces the number of
options from 4 to 2. I think if it's used in a span, then the content
of that span is a human-readable string identifying the subject, and
not a human-readable string identifying a second document about the
same subject. So that means that dcterms:subject is of type
subject-contents,object-sense.

If dcterms:subject is the property of a link, then i would say
probably... i don't know. it seems you probably want to have documents
sometimes be about senses, so then i would say
subject-contents,object-contents would be restrictive, so then i would
also interpret it as subject-contents,object-sense.

Unless other people would come to a different conclusion after reading
the definition, it's not broken. If they do, and there are examples
out there where people use dcterms:subject in any of the other three
ways, then yes, obviously it's broken, and the definition should be
clarified.

> Would not the URL have to be a typed literal in this case ?

that would be the equivalent of overloading a function to do different
things depending on the data type of the argument. i agree with you it
would certainly have been an option if we had four data types for
URIs. we don't, though. So each function that receives URIs should
just be clear about its meaning when you give it "a URI".

I think for dcterms:subject it's clear that it's
subject-contents,object-sense, and if an information provider uses it
in any of the other 3 ways, then i would expect information consumers
to choke on that and start to derive wrong conclusions.

Received on Monday, 14 May 2012 11:50:47 UTC