Re: RDF syntax 'improvements'?

Dan,

I generally agree with most of the opinions you offer, but have some other 
thoughts...

I believe the important part of RDF is its graph model.  The syntax is less 
crucial.

There are, I think, two primary purposes for an RDF serialization syntax:
(a) for interchange of pure RDF between applications, and
(b) for language-mixing RDF annotations into arbitrary XML documents.

I am not sure that there is any compelling reason, other than group 
history, to use XML for (a).  I suspect a simple triple-based format (or 
similar) would be easier to generate, parse and more compact to 
exchange.  I don't believe the simple structures of RDF graph elements gain 
much benefit from the power of XML.

Regarding case (b), there has been much talk of scraping RDF from existing 
XML documents, as an alternative to using an RDF-specific format for XML 
data.  This leads me to the thought that a meta-language describing a 
relationship between arbitrary XML and an RDF graph model might be more 
useful than a new RDF/XML serializaion format.  Consider:  formal syntax 
descriptions like BNF define a mapping between a linear text and a parse 
tree.  Why not a different kind of formal language to define a mapping 
between an XML tree and an RDF graph?

A "proof of pudding" for such a mapping language would be its ability to 
define a mapping for the current RDF syntax.

I note that DanC's work on using XSLT to map arbitrary XML to current RDF 
syntax would be one approach for such a project.  I'm not sufficiently 
familiar with XSLT, but I think I'd prefer a more declarative approach 
(might the XML QL work be used?).

#g
--


At 10:15 AM 8/4/00 -0400, Dan Brickley wrote:
>[disclaimer: this is my personal take and not any kind of W3C
>  pronouncement on the matter...]
>
>
>
>   Lee Jonas <lee.jonas@cakehouse.co.uk> asks:
> > Is it too soon after the RDF recommendation to suggest future
> > 'improvements'?  Should we just try to live with RDF as is?
>
>I have the long-rumoured 'RDF developer issues' list near the top of my
>todo list now. There are known bugs in the RDF syntax that need
>addressing. Quite how these are dealt with is something that we should be
>discussing in the RDF Interest Group. I've been going through the
>www-rdf-comments archive while preparing a report on the RDF Schema CR
>period; to pre-summarise this, one finding is that there are a few cases
>in which Schema, Model and Syntax issues become tangled. For example,
>primitive datatyping and RDF container constructs (Seq/Alt/Bag/etc), so
>summarising feedback on the RDF CR has involved summarising syntax etc
>issues too.
>
>
>I do not believe the RDF Model and Syntax errata document to reflect all
>the clear-cut 'syntax bugs' known to the community. Distinguishing
>clear-cut from 'up for discussion' is something we should be doing here. I
>intend to have some documents to fuel such discussion out within the
>month. We have had too much 'open issue churn' on this list. It would be
>good to put these issues behind us and get on with building things on top
>of the RDF core...
>
>
>Here are some things that could happen. We could/should bring the errata
>document for RDF M&S up to date with the experience of RDF implementors,
>and make available answers to FAQs where these seem clear, and writeup
>summaries for topics (eg. the xmlns prefix pairing business) that are
>perhaps not so clear. We could explore possibility of new work on a
>'better' rdf syntax, either as a W3C Working Group as an informal effort
>amongst RDF implementors on this list, with the intention of publishing
>either a new 'better syntax' REC (which would be a substantial piece of
>work) or an informational W3C Note outlining an alternative XML syntax
>for RDF models. ('we' being the RDF implementor community, ie. RDF IG)
>
>It could be that RDF Model and Syntax 1.0 plus cleaned up errata plus
>issue list and FAQ would remove much of the impetus for a Working Group
>activity in this area. At some point the errata needs to be rolled into a
>revision of the RDF Syntax document; this should be a largely mechanical
>affair. I don't think this is the same hypothetical work item as the
>ceation of a new alternate syntax for RDF (eg. the so-called 'by property'
>syntaxes that Sergey and TimBL have sketched).
>
>I'm not personally convinced that a new alternate RDF syntax is a
>priority right now, though I'd like to hear arguments to the contrary.
>The XSLT / Semantic Web Screenscraping threads on this
>list have shown how we can extract RDF models from all manner of well
>managed XML data. There are a fair number of RDF 1.0 parsers now, and
>significant effort has gone into creating these. I would rather see our
>time go on developing interoperability tests for these to get them up to
>production grade, learning through doing so about any grey areas in the
>syntax spec.
>
>In addition, XML Schema is not yet a REC; when we get such a thing,
>RDF syntaxes will be able to help themselves to more machinery
>(eg. datatyping, qnames in attribute values etc) than the original RDF
>Syntax WG had available. I do believe that we _will_ see some alternative
>strategies for encoding RDF-like data graphs in XML (eg. see the various
>papers from Andrew Layman and others, the SOAP submission etc); I'm not
>sure doing this as an RDF specific activity is a good use of
>resources. I'd rather see a good web data /graph syntax done for XML as a
>whole, and make sure this meets RDF's needs. Some have suggested XLink
>might even provide this; I'd like to see some feedback from implementors
>on feasibility of this.
>
>I'd also (and this is totally personal opinion as an RDF developer) rather
>see effort go into a clearly phrased discussion paper
>outlining the issues w.r.t. XML encoding of Web data graphs, summarising
>experience with deployment of RDF 1.0 syntax, contrast with alternate
>proposals (TimBL's, Sergey's, SOAP) for graphs-over-XML. Only once we had
>such a paper, something digestable by the mainstream XML world, would I
>feel comfortable proposing a major effort to define a better syntax for
>RDF. In other words, I'd rather go the route of working out a requirements
>document that says something of what RDF asks of a serialisation syntax,
>than to go immediately heads-down on defining some such syntax. My gut
>expectation is that after doing this, we'd realise that the thing we're
>after has a lot in common with non-RDF XML apps, and that a broader effort
>might make more sense.
>
>So... we should step back and ask for characterisations of what we
>want from an XML syntax for RDF. What are the must-haves?  What would the 
>goals
>be for any effort to provide a 'better' syntax? ie. what would make it
>better...? And then we need to ask who amongst us has time to commit to
>the projects sketched above. If the time, inclination and effort are
>there amongst W3C Members and the wider RDF world, then let's do it...
>
>IMHO etc.,
>
>Dan
>
>--
>mailto:danbri@w3.org

------------
Graham Klyne
(GK@ACM.ORG)

Received on Tuesday, 8 August 2000 03:42:39 UTC