- From: Rob Shearer <Rob.Shearer@networkinference.com>
- Date: Mon, 17 May 2004 18:23:37 -0700
- To: <kendall@monkeyfist.com>
- Cc: <public-rdf-dawg@w3.org>
Whatever "institutional reasons" you have for disliking SWRL, it really is a major W3C effort and singularly symbolizes the future growth of the semantic web stack. If you really don't think SWRL is relevent to the supplementation of RDF data with additional semantic knowledge then you and I clearly have very different views of what constitutes the semantic web 'layer cake'. As for other comments in this thread, we're not talking about source selection here. If you want to build an RDF graph by performing some kind of transitive closure or similar on another RDF graph, then query that graph entirely separately; it makes absolutely no sense to me that such graph transformations should be embedded within the query language. What is relevent to the query language is that you sometimes might have knowledge about a graph which can't be realized in that graph. If you know that one of two possible edges exists, but you don't know which one, then the queries "Must edge 1 exist?" and "Must edge 2 exist?" must return false, while "Does either edge 1 or edge 2 exist?" should be allowed to return true. I don't think the query language should dictate exactly how this knowledge should be obtained or encoded (although both OWL and SWRL offer ways of expressing it), but it is absolutely crucial that the query language not disallow the above operation, or we will have absolutely killed any possibility of a language which is relevent for true semantic querying. > -----Original Message----- > From: public-rdf-dawg-request@w3.org > [mailto:public-rdf-dawg-request@w3.org] On Behalf Of Kendall Clark > Sent: Monday, May 17, 2004 12:46 PM > To: Rob Shearer > Cc: public-rdf-dawg@w3.org > Subject: Re: I can accept... (Was: Re: Objective 4.6: > additional semantic knowledge) > > > > On Mon, May 17, 2004 at 12:04:52PM -0700, Rob Shearer wrote: > > > > The proposed rewording of my proposed rewording seems to be > nothing but > > a removal of the reference to SWRL. I included that > language in the list > > quite intentionally--I felt it was important to include a > language that > > can be used to describe RDF models but does NOT necessarily > have any RDF > > encoding. > > > > Why are we working so hard to make sure not to mention SWRL? > > That's language I can accept, given the way things are for me, > institutionally, as well as some of the semi-tech reasons I mentioned > in back channel mail. > > I can equally well ask what is so important about SWRL, which isn't a > standard and doesn't seem to have much behind it (?). > > So, I'll try again: I could accept language that replaced "SWRL" with > "or a language that can be used to describe RDF models but > which doesn't > have an RDF serialization" -- or words to that effect. That is, that's > my 2nd attempt to get what you say you want from "SWRL" w/out > mentioning SWRL. > > I'll point out, as a side note, that *no one in the entire known > universe* will read "and SWRL" to mean the things you want it to > mean. I.e., putting my editorial hat on over my UMD hat, the "and > SWRL" bit doesn't do what you want it to do, IMO, nearly as well as > explicit, generic language. > > Kendall Clark > >
Received on Monday, 17 May 2004 21:25:13 UTC