- From: Arnold deVos <adv@langdale.com.au>
- Date: Sat, 1 Jun 2002 16:25:24 +1000
- To: "Danny Ayers" <danny666@virgilio.it>, "Uche Ogbuji" <uche.ogbuji@fourthought.com>
- Cc: "Seaborne, Andy" <Andy_Seaborne@hplb.hpl.hp.com>, "'Patrick Stickler'" <patrick.stickler@nokia.com>, "ext Thomas B. Passin" <tpassin@comcast.net>, "RDF Interest" <www-rdf-interest@w3.org>
If anyone is interested in some fairly complete draft proposals for RDF query languages that are expressed.in RDF I have two: My first attempt [1] was posted here back in 2000. It is an XSLT-inspired template matching approach. It works at the level of the RDF graph, of course, not the document syntax. This was actually implemented and has been doing a good job in one of our products. More recently I posted [2] which attempts to turn DAML class expressions into queries. (Actually, the full regalia of DAML class expressions is not required if you don't like them.) This proposal has not been implemented so I can't claim to know whether it would be practical. In both proposals the query is expressed in RDF and applies to an RDF graph. But in [1] the result is abitrary XML, while in [2] the result is an RDF graph (a subset of the queried graph). In VERSA [3] I recall the result consists list structures (although Uche has pointed out to me that it could be a list of triples, effectively a graph or model). I find these differences in output just as significant as the form of the query language. Respectively, they are adapted to document production, further RDF processing and programatic manipulation. [1] http://www.langdale.com.au/RDF/NexusQueryLanguage.pdf [2] http://www.langdale.com.au/RDF/DAML-Query.html [3] http://uche.ogbuji.net/tech/rdf/versa - Arnold Arnold deVos Langdale Consultants http://www.langdale.com.au ----- Original Message ----- From: "Danny Ayers" <danny666@virgilio.it> To: "Uche Ogbuji" <uche.ogbuji@fourthought.com> Cc: "Seaborne, Andy" <Andy_Seaborne@hplb.hpl.hp.com>; "'Patrick Stickler'" <patrick.stickler@nokia.com>; "ext Thomas B. Passin" <tpassin@comcast.net>; "RDF Interest" <www-rdf-interest@w3.org> Sent: Saturday, June 01, 2002 8:57 AM Subject: RE: Innovation, community and queries > > > >> >properties(@"x:spam") > >> >@"x:spam" - properties() -> * > >> > >> Another good reason for an RDF QL in RDF! > > > >Possibly. It's hard to judge without seeing your proposal for > >expressing it > >in RDF instead. > > Fair response, and no, as you probably guessed, I don't have a proposal at > hand. > > >Of course, I must warn you I'm a sceptic. > > Me too, which is partly why I have a knee-jerk reaction to new syntaxes. > > >> Seriously though, I do think such a QL would be extremely > >useful, not only > >> because it would generally help interop. It would also mean that a whole > >> range of common expressions could become easier in RDF (without having to > >> drop into DAML-land), > > > >Examples? > > Essentially the kind of stuff like that which SQL scores on (almost > irrespective of the relational model) - e.g. forall kind of things. > > >> and also make things like XSLT-ish transformations a > >> lot more straightforward. > > > >We do this in 4Suite by using Versa to query and using XSLT itself > >to generate > >transformed RDF/XML. Works well, but we plan to come up with an > >XUpdate-like > >syntax as well. > > Hmm - I've experimented in the RDF+XSLT area myself, but have serious doubts > on its potential - ok, it can probably solve a lot of specific problems, but > having to think in trees is a bit ugly for the general case. > > >I'm not sure how Query in RDF would help make this more palatable > >than me. > >After all, the analog of RDF query in XSLT, XPath, is not in XML > >syntax. It > >still works quite well. > > Very true, but might it just be that with the DOM model doesn't need to be > good at metamodelling, which is something I would hope RDF languages would > be good at. > > >> Not unrelated to the interop point, the ability to > >> save sets of queries in a common format like RDF/XML has to be a > >plus - same > >> parser etc etc. > > > >This is a trivial matter of writing an RDF binding for whatever > >data model a > >QL uses. > > Writing a binding is trivial, writing a good binding is another matter. > > Cheers, > Danny. > >
Received on Saturday, 1 June 2002 14:25:27 UTC