[rdfweb-dev] Re: Styling schemas and *-a-matic

[Reto Bachmann-Gmuer]

-----BEGIN PGP SIGNED MESSAGE-----
Hash: SHA1

I've jusst experimenting with rendering schemas using trix:

http://www.wymiwyg.org/ontologies/knobot.trix -> the ontology as trix
with stylesheet
http://www.wymiwyg.org/ontologies/knobot -> the ontology in the flavour
preferred by your browser (probably html)
http://www.wymiwyg.org/ontologies/knobot.xml -> the ontology as rdf/xml,
with some sort of canonicalization (I have a priority list of properties
and types this allows rss to be understandable by some non-rdf aware
aggregators), but morten's concept is more convincing.

So what is faster and nicer, templates from trix of from canonicalized
rdf/xml (currently KnoBot just support both)?

cheers,
reto

Danny Ayers wrote:
| Morten Frederiksen wrote:
|
|> Hi Danny et al,
|>
|> On Thursday 29 April 2004 13:47, Danny Ayers wrote:
|>
|>
|>> I just came across your label vocab [1] - nice idea, I'll start using
|>> this asap.
|>>
|>
|> Thanks (I hope you'll use the "correct" namespace [1] for it, I've
|> seen you refer to it by its current URI).
|>
|>
|>
|>> But...would I be right in thinking that you used XSLT to generate the
|>> HTML from the schema? If so, how far have you got towards generalising
|>> this?
|>>
|>
|> You are indeed correct.
|>
|> It is far from finished, but it does show subclass trees etc. as it is
|> (up to three levels).
|>
|>
| Nice work!
|
|> I think the key point is the fact that I pre-normalise the RDF/XML by
|> using Redland/Raptor - parse it and reserialise, then group by subject
|> with XSLT (input [2], output [3]).
|>
|>
| Ah, I've been wondering about that.
|
|> Variations of this approach is how I now "always" consume RDF/XML
|> (also in FOAF Explorer), as it ends up in a predictable format. I
|> prefer this to alternate syntax proposals, as it's still valid
|> RDF/XML, even though it only uses a subset of the constructs. One of
|> the key points is not using the typed-elements construct - even though
|> it's nice for readability, it sucks when a node has multiple types...
|>
|>
| That hadn't even occurred to me.
|
|> When I intend to publicise the RDF/XML, I usually insert a step to
|> remove duplicate triples, and take another few steps by XSLT to make
|> it conform to the RSS 1.0 syntactic profile (including mod:content
|> syntax) and to indent it for readability.
|>
|>
| Right. A while back I was jumping through similar hoops with Jena,
| trying to get RSS 1.0-friendly (well, actually a bit more than that as
| one or two aggregators don't even seem to understand an rss: prefix).
|
|> This approach of course doesn't work for client-side XSLT, but I
|> figure that I might as well get the transformation done once, rather
|> than have everybody do it...
|>
|>
| Sounds reasonable ;-)
|
|>
|>
|>> It also occurred to me that it should be relatively straightforward to
|>> use XSLT to produce HTML forms, with fields generated from what appears
|>> in the schema.
|>>
|>
|> It's tricky, even though you can "guess" some instructions from
|> range/domain and OWL definitions (even though it's against the
|> principles of RDFS/OWL - they are for interpreting what you've said,
|> not telling you what to say).
|>
|>
| Another good point, though I'm hoping it'll be close to useful for
| simple, flattish stuff.
|
|> [1] http://purl.org/net/vocab/2004/03/label#
|> [2] http://www.wasab.dk/morten/2004/04/test.in.rdf
|> [3] http://www.wasab.dk/morten/2004/04/test.out.rdf
|>
|>
|>
| Thanks!
|
| Cheers,
| Danny.
|

-----BEGIN PGP SIGNATURE-----
Version: GnuPG v1.2.1 (Darwin)
Comment: Using GnuPG with Mozilla - http://enigmail.mozdev.org

iD8DBQFAk4sYD1pReGFYfq4RAo+uAKCLvs7aRysg+7xcBcJ4uo86SoXthQCfZh3M
L30QoB08gNzopdEULCJXR+4=
=4jQO
-----END PGP SIGNATURE-----