Re: Simpler syntax for RDF / Counter-proposal

Sergey,

I think you're on the right track.

Let me regurgitate to make sure I understand: the simple-counter-syntax for
RDF uses the namespace mechanism and only the namespace mechanism to
provide context for properties and classes. Right?
Let me explain what I'm thinking about.

It is very typical for documents to be locked into a particular
application-specific DTD. I would like to annotate the document using RDF.
Right now, I have to separate the RDF and the XML, which is un-natural.
What I'd like to see is a way to embed RDF in the XML so that a processor
could extract and strip the RDF annotation from the XML document.


So to use your syntax for annotation you end up not using the host XML tags
at all, as in

      <title>
        <dc:Title>My homepage</dc:Title>
      </title>

This is hard to work with for a number of reasons. For example, the host
DTD may have declared title to be PCDATA, and a document my require
validation for entity resolution. It would be a lot easier to have

      <title rdf:propertyType="<dc>#Title">
        My homepage
      </title>

This has an advantage that the propertyType could be declared as implicit
in a DTD; with careful surgery on a legacy DTD, millions of documents could
magically "become" RDF!

I know DTD's might be considered old, but let's not denigrate the virtues
of shipping software.

Eric


At 12:48 AM -0800 11/19/99, Sergey Melnik wrote:
>Tim,
>
>thanks for your clarifications.
>
>> Good.  But we should be incremental about this.
>
>I agree. Here is a syntax proposal for public discussion motivated by
>your draft:
>
>	http://www-db.stanford.edu/~melnik/rdf/syntax.html
>
>The proposal concentrates on the strawman's syntax. I did not consider
>the issues you raised in the Toolbox that would require extensions to
>the model/syntax. I believe higher-level languages should be built upon
>RDF, although the encoding might be verbose and inconvenient for humans.
>Freely quoting yourself in "Weaving the Web", RDF is a meta-model for
>constructing languages for the Web. In my view, the process of building
>such languages should start from the declarative languages of low
>expressiveness that have nice analytical properties and can be
>efficiently manipulated automatically. Examples include finite-state
>machines, specialized query translation languages, trust rules etc. that
>can be flexibly intertwined with each other. I believe, general-purpose
>logical expressions that have global validity are not the way to go, at
>least at the moment. But who knows: given your previous insights... ;)
>
>Sergey

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

Received on Friday, 19 November 1999 11:59:40 UTC