W3C home > Mailing lists > Public > www-rdf-interest@w3.org > November 1999

Re: Simpler syntax for RDF

From: Tim Berners-Lee <timbl@w3.org>
Date: Thu, 18 Nov 1999 09:46:14 -0500
Message-ID: <014e01bf31d3$aa565bb0$a60a1712@col.w3.org>
To: "Sergey Melnik" <melnik@DB.Stanford.EDU>
Cc: <www-rdf-interest@w3.org>

-----Original Message-----
From: Sergey Melnik <melnik@DB.Stanford.EDU>
To: Tim Berners-Lee <timbl@w3.org>
Cc: www-rdf-interest@w3.org <www-rdf-interest@w3.org>
Date: Tuesday, November 16, 1999 8:20 PM
Subject: Re: Simpler syntax for RDF


>Tim,
>
>in my view, there are two different perspectives to look at your
>proposal:
>
>1) on one hand, we need to be able to derive an RDF model from
>"arbitrary" XML (tough issue)


That is not what I am tackling.  To extract semantics from arbitrary XML
depends
on whether there is any rhyme or reason to it, and varies between a task for
a human being or an XSLT transform.

>2) on the other hand, a syntax that facilitates exchange of RDF between
>applications is needed that is easy to understand, implement etc.

Well, useful it is, and your <triple> is equivalent to the <rdf:property>

which I threw in for exactly that purpose.

But the motivation for the unstriped syntax was more the motivation
for the orginal M&S 1.0

3) We need to be able to make a syntax which will allow an application whose
data in fact can be expressed reasonably in an RDF model to serialize it
so that any RDF-aware application will be able to recognize it in within
an XML document and extract the RDF graph.  I would leave open a way of
doingthis
which requires reading the schema because I haven't learned enough about the
hooks
for making the necessary declations in xml-schema langauge.  I focussed on a
syntax
which would put minium of flags like "rdf:about" into the documentwhich
would let the parse
extract the tree *without* the schema.


>Let me start with (2) since this is the simple point. My experience
>during contributing to SiRPAC and implementing serialization for RDF
>suggests, that for many cases a "*poor* strawman's syntax" would be just
>fine:
>
><triple subject="URI" predicate="URI" object="URI"/>
><triple subject="URI" predicate="URI">literal</triple>
>
>A "desperate Perl hacker" (remember this paper?) could implement in
>within minutes. You still have all the advantages of Unicode-based XML
>etc.


Agreed. (Aren't all perl hackers desperate? ;-)

>As to (1):


I will see whether these remarks apply to the need (3) ...

>there are two big problems that I see, and a number of smaller ones. The
>big problems are:
>
>(a) since parsing relies on the schema information about XML tags,
>parsing without schema would not be possible (I might have misunderstood
>this). In that case, however, we're moving to where we came from: SGML
>is not parseable without DTDs.


The intent is that attributes can be inserted to make schema unnecessary.

>(b) default context: in the current specs, the context is limited to one
>level within the Description. Higher context-sensitivity (many-level) is
>problematic if one needs to extract information out of context. Even in
>case of nested RDF/XML you'll find just a couple of RDF pages out there,
>that use it correctly.


In the current specs, contenxt is carried down all the way, alternatibg
between
node typeand arc type.

In general, you cannot extract information out of context unless you are
somehow
told you can.  For example

<foo:bar>
   <rdf:property pname="http://dc.org/dc1#author"
   rdf:for="theBook"
   rdf:value="Ralph">
</foo:bar>

does not tell you anything unless you know something about foo:bar.
The easiest thing is to know that it is what I call "RDF transparent".
But you can't assume it -- or I can never define <not>, for example.


>You have made a "call for implementation". However, so many issues still
>have to be resolved, that I (as a potential implementor) would just not
>know where to start.


Ok, don't then. It is at the hack and see whether the spec makes sense
stage, not the  well-maintained body of open source code stage.

Tim

(rest of message about M&S 1.0 replied to separately)

_____________________________________________________

>
>Before such issues have been resolved, I'd be very careful about
>motivating people to implement something. Currently I am a courtesy
>maintainer of SiRPAC. However, to be frank, I'm not willing to
>rewrite/adapt it as long as the specs are flawed.

I agree that things have to be thought out first.

>I like the spirit of your proposal. Still, I think that an organized
>effort is needed to produce something targeted at RDF M&S 2.0. And I'd
>definitely love to contribute to it.

Good.  But we should be incremental about this.

>Best,
>Sergey
>
Received on Thursday, 18 November 1999 09:44:59 GMT

This archive was generated by hypermail 2.2.0+W3C-0.50 : Monday, 7 December 2009 10:51:42 GMT