Re: RX - another RDF in XML format [WAS: Re: a simpler form of rdf xml]

Heh! Really interesting. Thanks. :-)

This makes me want to see if I can formalize the serialization format I 
was proposing. As with yours and with the really interesting proposal 
by Sergy
Melnik [1], (and it turns out Tim Berners Lee [2]!) I was proposing 
that all
xml elements be links. Cool.

==========DO NOT READ BELOW. THOUGHT UNDER CONSTRUCTION=============
===============CURRENTLY BROKEN=====================================

But I think mine is both a lot simpler and a lot more general. Because 
all I do is
add that xml element attributes be treated the same way.
   - The subject of the attribute relation is the object of its element's
     relation object.
   - The object of the attribute relation is a blank node _attObj
   - The blankNode _attObj is related to the value of the attribute by an
     rdf:string relation.

    <entry href="http://bblfish.net/blog/">

    is represented as
    _something :entry _entry_obj .
    _entry_obj :href _href_obj .
    _href_obj rdf:string "http://bblfish.net/blog" .

And I apply the same principle to the xml elements text content, if it 
has any:
The text is related to the object of the containing xml node's relation 
by an
rdf:string relationship.

     <entry href="http://bblfish.net/blog/">
     some text
     </entry>

     is represented as
     _entry_obj rdf:string "some text"

And that is all I specify.

----------------------That'a All------------------------

All the rest is to be defined in an ontology.

1. The RX format
----------------

So I think your transformation just turns out to be special case of 
what I
have specified. Namely one in which you have limited for example all the
attributes to only have uris as values.


2. TimBl's proposal
-------------------

This is how we can now explain TimBl's [2] rdf:for, rdf:about, ...

  - rdf:about is an identity property. It relates a node to itself. It is
   functional, inverse functional, symmetric and transitive. *Any* other
   such property would do equally well.

   <frontm rdf:about="uri:urn:theBook">The Varieties of 
Reference</frontm>

   has the following

   _subj :frontm  _obj .
                  _obj rdf:about _about .
                                 _about rdf:string "uri:urn:theBook" .
                  _obj rdf:string "The Varieties of Reference"

   That's all!

   Now because rdf:about is an identity relation the above graph is 
equivalent
   to one where _obj is replaced with _about, ie:

   _subj :frontm  _about .
                  _about rdf:string "uri:urn:theBook" .
                  _about rdf:string "The Varieties of Reference" .

   !OOPS! This does not work. Because we have two rdf:string relations 
on about.
And I wanted to conclude that because

   _about rdf:type <xxx:anyUri> .

   that we could deduce that _about == <uri:urn:theBook>

   So my proposal does create a problem here... Perhaps I have to change
  the way I relate _about to the string content of the frontm xml tag. 
Perhaps
I have to create a rdf:stringContent relation between _about and
"The Varieties of Reference" so that the xml turns out to be:

  _subj :frontm  _about .
                 _about rdf:string "uri:urn:theBook" .
                 _about rdf:stringContent "The Varieties of Reference" .

and perhaps we just cannot deduce from

    _about rdf:stringContent "The Varieties of Reference"
    _about rdf:type <rdf:anyUri>

  that

   _about == <the Varieties of Reference>

which would be good.

but then that would limit one of the conclusions I thought I could draw
with constructs such as

    <id>uri:urn:theBook</id>

which I was hoping would turn out into

    _subj :id _obj .
              _obj rdf:stringContent "uri:urn:theBook" .

	and knowledge that

	_obj rdf:type xxx:anyUri .

     to conclude that

     _subj :id  <uri:urn:theBook> .




On 17 Jan 2005, at 23:46, Josh Sled wrote:

>
> I wrote up in a bit more detail some of the ideas I was jabbering about
> last week.
>
> http://asynchronous.org/rx/
>
> It's yet-another RDF-in-XML format.  In short, it turned out to be
> nearly identical to Sergey Melnik's Simplified Syntax for RDF [1], 
> which
> I had not seen before.
>
> I've briefly described how RX compares to the many other RDF-in-XML
> formats along the same [and different] lines, which was quite
> interesting in and of itself.
>
> ...jsled
>
> [1] http://www-db.stanford.edu/~melnik/rdf/syntax.html
[2] http://www.w3.org/DesignIssues/Syntax

Received on Tuesday, 18 January 2005 13:24:59 UTC