Re: Simpler syntax for RDF

I found this to be very interesting, and also the responses from Sergey,
Stefan and others.

I think that TimBL (and the Cambridge [communique]) is trying to address
the fact that for those of us who are unschooled in metadata and knowledge
representation, RDF is difficult to use. The different syntaxes make it
very easy to make mistakes when composing it by hand, and I agree that it
is the "striping" which makes it hard. Thus, while RDF is not hard to
understand, it's hard to "speak".

However, I don't find that TimBL's "simpler" syntax is a whole lot easier.
I agree with Sergey Melnik's analysis of the problem. His atomic syntax IS
very easy.

Perhaps if the goal of "simple" XML were less ambitious than a complete
expression of RDF, and scaled back to simple mark-up, then Sergey's syntax
could be almost directly applied.

I find that XML lends itself quite nicely to representing data structures.
A goal of "RDF Markup" might be to avoid any alteration of the XML tree
structure. It would then be easy to decouple the RDF from the XML.

My guess is that in the majority of cases, rdf markup of xml is a matter of
attaching simple self-referential statements to elements.

The relevant statements to put on an "element" would be

  (subject = this )
<myTag rdfm:predicate="URI" rdfm:object="URI"/>
<myTag rdfm:subject="URI" rdfm:predicate="URI">literal</myTag>

(predicate = this )
<myTag rdfm:subject="URI" rdfm:object="URI"/>
<myTag rdfm:subject="URI">literal</myTag>

(object= this)
<myTag rdfm:subject="URI" rdfm:predicate="URI">literal</myTag>

URI could include "parent" and "child" (I'm assuming that there is an xpath
or xpointer to make these into URI?)

This should be trivial to parse, as each statement is strictly local to an
element.




At 6:12 PM -0500 11/16/99, Tim Berners-Lee wrote:
>The Cambridge [communique] meeting had a consensus that a simpler syntax for
>RDF. At the time I mentioned that I had done some thinking about it. The
>results are at
>
>http://www.w3.org/DesignIssues/Syntax
>
>Comments welcome.
>
>The basic idea is that XML elements represent RDF properties: there is
>no "striping" as in the standard RDF syntax, in which alternate layers in
>the nesting
>represent node types and arc types.  In the "unstriped" syntax,
>node types (if used) can be made explicit by an arc, or can be deduced from
>the domain and range of properties.
>
>I have found that many applications seem to look (to me!!) much more
>obvious in this syntax.
>
>The document is not totally baked (say where it doesn't make sense)
>but I think the next step should be code.
>I would love to be able to find time to do it myself, but help is of course
>useful.
>
>Things which I was considering were
>
>- an XSLT transform from unstriped to striped syntax (and/or back).
>- a modified RDF parser in your favorite language which accepts either
>current or unstriped syntax.
>
>The page does not have examples - but the syntax is used in some
>accompanying
>pages such as http://www.w3.org/DesignIssues/Toolbox.html
>
>Tim Berners-Lee
>no hat

Eric Hellman
Openly Informatics, Inc.
http://www.openly.com/           21st Century Information Infrastructure

Received on Wednesday, 17 November 1999 15:51:05 UTC