RE: Again: Anonymous Resources

[caught in spam trap -rrs]

Date: Mon, 12 Mar 2001 09:37:42 -0500 (EST)
From: "Lee Jonas" <ljonas@acm.org>
To: <www-rdf-interest@w3.org>, <love26@gorge.net>
Message-ID: <PJEHIGCEAMAHEAPOOFBEIEBCCBAA.ljonas@acm.org>
Subject: RE: Again: Anonymous Resources


love26@gorge.net (William Loughborough) wrote:
>
>At 04:56 PM 3/9/01 +0000, Lee Jonas wrote:
>>basic essence of that animal doesn't change no matter how you refer to it.
>
>As in "sticks and stones may break my bones but names will never hurt me"?
>
>We (and any other resources") are basically affected (even to our "basic
>essence") by means/content of reference. I fail to see how "assigning a
>handle to it" is different from "naming" it.
>
>--
>Love.
>                 ACCESSIBILITY IS RIGHT - NOT PRIVILEGE

Would you be any different if I refer to you as love26@gorge.net instead of
William?

We could argue metaphysics until we are blue in the face.  The truth of the
matter is that either stance is justifiable.  The real crux of it is what
impact the implications of each stance has on RDF and its use.

RDF constructs describe resources.  If those mere descriptions automatically
assign permanent URI-references to those resources then there can *never* be
anonymous resources in RDF.  This would certainly be one way for the RDF2
spec to go.

However, there are a number of disadvantages:

* You would have to insist that an application names every resource -
including those 'convenience resources' for associating qualifiers with
underlying values.

* Alternatively, all parsers must have a parser-independent way of
consistently reproducing the same global, unique, permanent identifier for
resources described in an XML document without a URI-reference explicitly
assigned.

IMHO, the first of these is unacceptable whilst the second is extremely
difficult.

The second option (keeping the concept of anonymous resources in RDF2) also
has implications:  Temporary identifiers must be assigned to them during
parsing.

Although this can also be viewed at a superficial level as *naming*
anonymous resources, it can more acurately be viewed as bootstrapping XML,
N3, SemEnglish, etc. into more abstract RDF model constructs (e.g. instances
of Java classes) that are not associated with a particular URI-reference.

Again IMHO, serialisation is only of interest to RDF parser developers, not
RDF application developers.  RDF applications can fully utilise anonymous
resources as intended in the current RDF spec, and parsers are free to use
far simpler identifier generation schemes, as these ids don't have to be
reproducable by them (or any other parser) in a globally unique fashion.

Regards

Lee

Received on Monday, 12 March 2001 11:36:49 UTC