- From: Jeremy Carroll <jjc@hpl.hp.com>
- Date: Mon, 04 Jun 2007 14:48:10 +0100
- To: Matt Williams <matthew.williams@cancer.org.uk>
- CC: Taowei David Wang <tw7@cs.umd.edu>, Ulrike Sattler <sattler@cs.man.ac.uk>, Ian Horrocks <horrocks@cs.man.ac.uk>, Owl Dev <public-owl-dev@w3.org>, Semantic Web <semantic-web@w3.org>, Bijan Parsia <bparsia@cs.man.ac.uk>
Matt Williams wrote: >(I suspect they're > different - the rdf graph is not, I hope, infinite). > In the Jena implementation, the rdf graph is allowed to be infinite, we believe this to be standards conformant. It is, I believe, possible, in Jena, to do the following sorts of steps: a) construct a finite RDF graph G (e.g. by reading a file) b) construct an object that corresponds to the infinite graph R(G) corresponding to applying a finite rule set R to G (from (a)). c) query the infinite graph with a finite query Q, to get an object corresponding to the query answer Q(R(G)). If this is finite, and modulo some unspecified tractability constraints, it is then possible to turn this query answer into a more concrete representation of a finite set. If it is infinite, it is possible to process a finite number of its members, (in an order of our choosing, I believe). Having said this, I think the discussion in this thread is interesting, and may be useful to help us provide mechanisms to ensure that Q(R(G)) is finite. Jeremy -- Hewlett-Packard Limited registered Office: Cain Road, Bracknell, Berks RG12 1HN Registered No: 690597 England
Received on Monday, 4 June 2007 13:48:32 UTC