W3C home > Mailing lists > Public > www-rdf-interest@w3.org > September 2000

FW: RE: aboutEachPrefix: drop or specify via startsWith

From: David Wagner <dwagner@sa.kevric.com>
Date: Fri, 1 Sep 2000 14:30:24 -0500
To: <www-rdf-interest@w3.org>
Message-ID: <C9B23C6D1899D21188F600C00D009F86254D02@KEVRICSA1>
[I think I sent this to the wrong address first.  Please pardon if you
get it twice.]

I noticed another serious problem with aboutEachPrefix when reversing
the RDF concept to generate a website (structure and content) from a
single RDF linkbase.  If I'm not mistaken, the intent of the
aboutEachPrefix referent is to make statements about resources based
on a URI pattern, but the current definition,

 "The members of this Bag container are all the resources
  whose absolute form resource identifiers begin with the
  character string given as the value of aboutEachPrefix."

seems to be completely inadequate.  It is based on the assumption that
URIs represent a hierarchical structure from left to right, and this
is often not the case, nor do I think it could always be so in every
domain.

The most obvious example is that of subdomains.  If I wish to assert
something about all the resources at example.com,
aboutEachPrefix="http://example.com" would not work since it should
also apply to foo.example.com, bar.example.com, and any other
subdomain example.com wished to add in the future.  As far as I can
tell, I cannot do this with RDF, and it is critical for such things as
site content rating.

Also, Even if we could all agree on a standard URI naming convention
(no need to get into that here), the arbitrary order of orthogonal
portions of an identifier should not limit our ability to make
statements based on a URI pattern.  If orthogonal parameters are
mapped to the URI, the order in which they appear would be chosen
arbitrarily, and not necessarily in the order convenient for
statements someone may like to make.  Say example.org has archives of
news and weather by year;  aboutEachPrefix only allows statements
based on the order in which example.com decided to put these, limiting
drastically our ability to say something about the content at these
locations.

http://example.org/1999/news    http://example.org/1999/weather
http://example.org/news/1999    http://example.org/weather/1999

Going further in, when I first saw RDFM&S in draft many moons ago I
assumed it would (as other W3C documents) be updated on the next
revision to include current work in other areas where applicable.  The
aboutEachPrefix referent seemed obvious to at least allow for use of
an XPath or XPointer construction to allow statements about portions
of a document.  I was surprised to see there has been little
discussion of this.

In the archives I did find
http://lists.w3.org/Archives/Public/www-rdf-comments/1998OctDec/0026.h
tml. (Comment 4 applies.)  Since the W3C does not wish to specify a
genuinely useful construct (say, aboutEachPattern), and the current
definition opens the model and causes as much difficulty as any more
complicated and useful definition, imvho, I think it would be better
to drop aboutEachPrefix entirely.

These are two cents from --David.
Received on Friday, 1 September 2000 15:38:44 GMT

This archive was generated by hypermail 2.2.0+W3C-0.50 : Monday, 7 December 2009 10:51:44 GMT