- From: Brian McBride <bwm@hplb.hpl.hp.com>
- Date: Thu, 24 Apr 2003 18:02:29 +0100
- To: "Peter F. Patel-Schneider" <pfps@research.bell-labs.com>, jjc@hplb.hpl.hp.com, danc@w3.org
- Cc: danbri@w3.org, dave.beckett@bristol.ac.uk, w3c-rdfcore-wg@w3.org
At 07:47 24/04/2003 -0400, Peter F. Patel-Schneider wrote: [...] > > [The proposal is that RDF/XML parsers should not emit triples > > _:x rdf:type rdf:List . > > since they are redundant). > >I oppose this change. OWL already depends on them being there. That seems pretty clear cut - by the time the RDF gets to an Owl (lite or DL) processor the rdf:type triples need to be there. But that's processing model talk. So, forgive me talking processing model a little - then I'll get back on message. We could envisage a processing chain where RDF gets fed through some RDFS based inference process before going into an Owl processor to put the rdf:type triples back in, if they were omitted by a parser. Part of what Jeremy calls an Owl Tidy. More simply we could envisage a parser that had a switch to control whether these rdf:type triples were generated or not. The latter of these seems an appropriate pragmatic approach, if the WG felt that the presence of the rdf:type triples is a problem. Whilst we don't talk about processing model we might add a note to the syntax doc saying something like: [[ Note: The (x, rdf:type, rdf:List) triples are required for some applications, e.g. Owl lite and DL processors, but may be redundant and unnecessary for some applications. Implementations may provide a facility not to generate these triples where there presence would be unhelpful. ]] Whether or not we included such a note we could response to Tim along the lines of: [[ The RDFCore WG has considered your last call comment URL and resolved URL to reject it on the grounds that Owl Lite and DL relies on the presence of these triples. The WG notes however, that there is nothing to preclude an implementation from not generating them in circumstances where they are not needed. Further, a general purpose RDF processor, such as a parser, could be built in such a way that the generation of these triples is configurable, and there is no requirement that the default configuration is that they should be generated. Possible words about adding note to the document, if the WG decides to adopt it. ]]
Received on Thursday, 24 April 2003 13:01:57 UTC