- From: Graham Klyne <GK@ninebynine.org>
- Date: Tue, 11 May 2004 17:32:33 +0100
- To: Damian Steer <damste@hplb.hpl.hp.com>
- Cc: RDF interest group <www-rdf-interest@w3.org>
At 11:57 11/05/04 +0100, Damian Steer wrote: >| Simple RDF query processing is not hard to do, so this might prove to be >| a simple and flexible approach to providing a standard way to mesh >| content adaptation with client capability descriptions. Inference is a >| little harder, but can be reasonably easily to express in a functional >| language (which XSLT is) given the right query primitives, so this >| doesn't look like a vast leap beyond currently exists; I think most of >| the details designs already exist. > >Treehugger just operates over the jena inferred graph (which has its own >issues). Personally I wouldn't fancy doing eg subclass/property closure >in XSLT (XQuery perhaps) but it's certainly possible, if messy. Ah, yes, recursive query processing (closure) might be more challenging using XSLT. (I'm not convinced of this yet, but I haven't thought enough about the details.) But I think the CC/PP default processing could possibly be handled by a simpler form of rule application. I would expect to keep the XPath design details for accessing RDF graph data separate from the rule application logic. But there are many things I haven't thought through, so this is still at the level of "thinking aloud". #g ------------ Graham Klyne For email: http://www.ninebynine.org/#Contact
Received on Tuesday, 11 May 2004 12:40:47 UTC