Re: [office-metadata] Any option for a RDF/XML subset?

[Bruce D'Arcus <bdarcus@gmail.com>]

On Feb 14, 2007, at 3:02 AM, Michael Brauer wrote:

> Regarding the constraints itself: My understanding is that RDF-XML 
> supports multiple variants. Unfortunately, I have no experience
> - how difficult it is to implement all of them,
> - what RDF-XML frameworks exists, and whether they support all 
> variants,
> - whether there is a common subset of RDF-XML that is known to work in 
> all applications.

There aren't multiple variants per se. There is RDF/XML, the W3C 
standard. It has a notoriously flexible syntax. For example, properties 
may be represented as attributes or elements. There are other examples 
like this.

There are also some RDF-XML profiles (subsets). XMP is an example 
(though oddly they subset the model rather than per se the syntax). RSS 
1.0 is another.

So your question about "multiple variants" and implementing "all" of 
them can't really be answered (because there are not really multiple 
discrete things), except to say best to support RDF/XML if you want to 
be fully compatible.

The frameworks/libraries (Redland, Jena, RDFLib, etc.) will of course 
handle reading and writing the RDF/XML for you. You work with the model 
through the API. This is indeed the point of using RDF as a common 
model.

> What I'm mostly concerned is the implementability of metadata in ODF. 
> If it is easy to implement the full RDF-XML specification, then we may 
> not need any constrains.

The really easy approach is to just use some existing open source RDF 
library, which will have all that support. I'm not sure how hard it is 
to implement an RDF/XML parser if you needed to do, but there's plenty 
of examples around in existing implementations.

The interesting issue is that if you constrain the syntax, you might 
make those RDF tools work less well. E.g. there is a trade-off between 
friendliness to, say, XPath, and to RDF.

Bruce