Re: Ed Summers blog post on 'chimera' approach

Jonathan Rees wrote:
> The purpose of section 5.5 is to give a fair and balanced exposition
> of the idea given here:
> 
> http://inkdroid.org/journal/2010/07/07/linking-things-and-common-sense/
> 
> Our document ought to be readable by this audience. Perhaps not easily
> readable, but it shouldn't be unnecessarily alienating. Ideally, it
> should make Ed and others feel that they have at least been heard.
> 
> If I understand Harry's comment about there only being a problem with
> 200 in the presence of owl:sameAs, then I think Harry is espousing a
> similar view. But I could be wrong.
> 
> This post and its comments also provide a good list of misconceptions
> that we might try to straighten out in the document (although some of
> them seem to be beyond redemption).

Yes, what Ed outlines is perfectly reasonable in many situations.

It appears that Ed is simply saying that if one asserts:

   <http://en.wikipedia.org/wiki/William_Shakespeares> dcterms:modified 
"2010-06-28T17:02:41-04:00"^^<http://www.w3.org/2001/XMLSchema#dateTime>

then one can infer:

  Document(<http://en.wikipedia.org/wiki/William_Shakespeares>)

and if one also asserts:

  <http://en.wikipedia.org/wiki/William_Shakespeare> a foaf:Person ; 
foaf:name "William Shakespeare" .

then one can infer (well read):

   Person(<http://en.wikipedia.org/wiki/William_Shakespeare>)

and that properties could be classified as relating to the classified 
object, such that you'd have (replacing 
<http://en.wikipedia.org/wiki/William_Shakespeare> for <ws> below):

   Person(<ws>) foaf:name "William Shakespeare" .
   Document(<ws>) cc:license 
<http://creativecommons.org/licenses/by-sa/3.0/> .

and:
   Person(<ws>) :differentFrom Document(<ws>) .

This appears to be Many Sorted FOL, or Typed FOL with multiple domains 
of discourse, something RDF is not (from my read anyway!).

Even if it were though, we'd still run in to problems, and the case Ed 
doesn't cover, which is for example:

   <http://en.wikipedia.org/wiki/Frankenstein>

In this case, when it's describing something that is itself a 
document/book, the whole thing falls apart (you'd have two authors, two 
creation dates, etc), which I'm sure you'll agree is a bit of a monster.

Hence, I have to disagree w/ Ed's "5. If it’s not clear, maybe the 
vocabulary sucks."

Thing is though, it is a reasonable proposal in a decent portion of use 
cases, and combined with the "I won't describe the document" and "I'll 
refer to the document by some other URI" voices, it does cover a fair 
majority of the cases, but only in certain domains, and only by having 
some special rule for metadata that lets you override the norm to say 
"no it's not a doc, this over here is the doc" (like the CL approach). 
At web scale though, it's far from an ideal solution, and really doesn't 
help the worlds Librarians, or the Frankenstein case.

Best,

Nathan

Received on Tuesday, 5 April 2011 13:41:23 UTC