(unknown charset) Re: Applicability of the RDF formal grammer

Hello Victor.

I'm not a RDF expert so please take this comment as an advice from an application
developer not as definitive answer to you question.

I think the RDF file shown in the Appendix A of the 'RDF Schema Specification'
follows the RDF/XML grammar. In the RDF/XML BNF, production for
"description" actually allows typedNode, and descriptions in the example RDF matches
this construct.

Your second question is related to the third abbreviated syntax of RDF/XML.
Please refer to section 2.2.2 of [RDF M & S] again.

I think the key to understand [RDF M & S] is that clearly separating abstract RDF model
from the syntax which is formally called RDF/XML. Because two deferent things are specified
in the same document, they are sometimes confused with each other, though.

RDF model is defined in section 5 of [RDF M & S] using several propositions
and the model, I think, is quite simple. (It's just equivalent to the directed arc and node graph.)

On the other hand, syntax specified in [RDF M & S] is just "one" way of serializing the
abstract RDF model for data exchange. As mentioned in the first section of the [RDF M & S],
in theory, there may be different way of serializing the RDF model, so the same RDF model
may have different representations in serialized data form. For example, in WAP UAProf, which I'm
now targeting, data is transmitted in WBXML(binary encoded XML) form for data compression
rather than usual text XML.

Considering the independence between model and syntax of RDF, the flexibility for
having various kind of abbreviated syntax is possible.

So the following two RDF/XML data is quite different in each serialized form, the RDF
model produced (such as SiRPAC's output) is the same.

<rdfs:Class rdf:ID="MotorVehicle">
    ...
</rdfs:Class>
...

and

<rdf:Description ID="MotorVehicle">
  <rdf:type resource="http://www.w3.org/2000/01/rdf-schema#Class"/>
  ...
</rdf:Description>

I think that it is the most difficult thing to grasp that one RDF model can have many different
form of data. (Despite the fact that I'm not good at using English, I have to read more than 3
times before I can finally grasp the main point.)

Personally I don't like such flexibility because It will make implementation of validating
parser such as SiRPAC much more difficult.

As a programmer I prefer XP's way of simplicity rather than unnecessary flexibility.
But not only in the object oriented world but also in other areas of disciplines such as
math or physics, separation of the model and representations is quite often used.

/*************************************
 *
 *   Ryo Asai  asai@sec.co.jp
 *
 *  Systems Engineering Consultants Co.,Ltd.
 *  Marketing department
 *
 *************************************/

----- Original Message -----
From: Victor LIU
To: www-rdf-comments@w3.org
Sent: Sunday, December 10, 2000 5:15 PM
Subject: Applicability of the RDF formal grammer


Dear W3C Members:

I've just read through the documents on RDF model and schema.  There is one question confusing me
right now.  In the 'RDF Model and Syntax Specification', section 6 gives the formal grammar for RDF.
From the BNF rules, a RDF file is defined as a list of 'obj', where each 'obj' is either a
description or a container.  However, some real RDF files, such as the one shown in Appendix A of
the 'RDF Schema Specification', do not follow this basic construct.  Does this mean that the BNF
syntax in section 6 will not apply to certain RDF files?

Another question:
What's the difference between:
...
<rdfs:Class rdf:ID="MotorVehicle">
    ...
</rdfs:Class>
...

and

<rdf:Description ID="MotorVehicle">
  <rdf:type resource="http://www.w3.org/2000/01/rdf-schema#Class"/>
  ...
</rdf:Description>

Thanks a lot

Victor Liu
UCLA Computer Science Dept.

Received on Tuesday, 12 December 2000 01:52:32 UTC