- From: Jeremy Carroll <jjc@hplb.hpl.hp.com>
- Date: Wed, 08 Dec 2004 15:29:03 +0000
- To: "Eric Prud'hommeaux" <eric@w3.org>, "Seaborne, Andy" <andy.seaborne@hp.com>
- CC: "www-archive@w3.org" <www-archive@w3.org>
Hi Eric I spent a day thinking about subgraph isomorphism last week. The results were largely negative, and apply to the SPARQL testing question you asked me a while ago. In as much as SW concerns are about semantics, the syntactic operations involved in (sub)graph isomorphism are inappropriate. For query language issues this is shown most clearly in a triple not getting used up if it matches. e.g. comparing <a> <b> <c> . <a> <b> _:d . with <a> <b> <c> . _:e <b> <c> . these two mutually entail one another but have no sub-graph isomorphism relationship between them. This sort of example can easily be mapped into QL concerns (e.g. adding a ?x node into one of the graphs), or thinking of the graph as a query result in some way. Sorry, I think this is a blank. However, Herman ter Horst's work on RDF(S) entailment does show that it is plausible to build a SPARQL test suite in terms of simple entailment. (See Herman's ISWC paper) Jeremy
Received on Wednesday, 8 December 2004 15:29:34 UTC