RE: HTTP mechanics +1, IR semantics -1

Hello Harry, Alan,

> -----Original Message-----
> From: public-awwsw-request@w3.org
> [mailto:public-awwsw-request@w3.org] On Behalf Of Harry Halpin
> Sent: 14 July 2008 22:36
> To: Alan Ruttenberg
> Cc: public-awwsw@w3.org; Rees Jonathan
> Subject: Re: HTTP mechanics +1, IR semantics -1
>
>
> Alan Ruttenberg wrote:
> >
> > As stated, this is not the activity that I signed on to
> participate in.
> >
> > Having a clear and well specified architecture based on sensible and
> > understandable documentation, that has the potential number to server
> > development in a wide variety of circumstances, including scientific
> > ones, is of interest.
>
> Yes, this of interest. Serving a particular application, and by implication,
> suggesting that it is the canonical semantic web application, does not.

/me too.

> However, if we want to actually merge data across applications, having
> some sort of provenance ontology (did we GRDDL to get this RDF? Did we
> use RDFa?

Interesting... I'm beginning to turn most of my attention to the topic of provenance on the semantic web.

FWIW: I've alway's had a bit of a problem wrt to RFDa and GRDDL wrt to media-types and intentions. Both RDFa and GRDDL enable 'triples' to be 'extracted' from (webarch:representations of) document (like-things). What is *much* less clear is whether its was the intent of the orginal author to incant such triples. So... giving an account of how some set of triples happened to be asserted in some RDF model/OWL A-box whatever, seems vital rather than simply attributing them without qualification to the document author.

I haven't properly looked at GRDDL (and to some extent RDFa) processing, but I am bothered (in the sense that whilst it may be possible to get it right - I suspect that without care that it can be got horribly wrong) establishing correct subject URIs particularly in the presense of relative URI references and implicit/explicit base URI (plus media-type issues over the variant media types that may be associated with a reference).

> Did we have to run it through some data-munging? Manually
> replace some URIs in a pipeline?) is absolutely a second crucial
> component of the architecture of the Semantic Web, once we have a basic
> normative algorithm for retrieving SemWeb data given a URI. To me, it
> appears that documentation URIs/information resources would be
> components of such a provenance ontology.
>
>           take care,
>              harry

Stuart
--
Hewlett-Packard Limited registered Office: Cain Road, Bracknell, Berks RG12 1HN
Registered No: 690597 England

Received on Tuesday, 15 July 2008 11:43:05 UTC