W3C home > Mailing lists > Public > public-owl-wg@w3.org > May 2008

Re: [FULL] ISSUE-119: the other approach

From: Jeremy Carroll <jjc@hpl.hp.com>
Date: Wed, 21 May 2008 11:10:49 +0100
Message-ID: <4833F529.5090002@hpl.hp.com>
To: Ian Horrocks <ian.horrocks@comlab.ox.ac.uk>
CC: Michael Schneider <schneid@fzi.de>, "Web Ontology Language ((OWL)) Working Group WG" <public-owl-wg@w3.org>

Ian Horrocks wrote:
> I talked to Peter about this and we agreed that Jeremy's proposal looks 
> promising.

Please see the three atatchments (Carroll & Turner) to

We show that OWL1 Full without the comprehension principles is 
consistent, and does not break most RDF graphs that do not use the OWL 
vocabulary. We discuss the role of the comprehension principles in OWL 
semantics, and how to maintain the relationship between OWL Full and OWL 
DL by reinterpreting the comprehension principles as permitted steps 
when checking an entailment, rather than as model theoretic principles 
constraining the universe of interpretation. Starting with such a graph 
we build a Herbrand model, using, amongst other things, an RDFS ruleset, 
and syntactic analogs of the semantic “if and only if” conditions on the 
RDFS and OWL vocabulary. The ordering of these steps is carefully 
chosen, along with some initialization data, to break the cyclic 
dependencies between the various conditions. The normal Herbrand 
interpretation of this graph as its own model then suffices.
The main result follows by using an empty graph in this construction.
We discuss the relevance of our results, both to OWL2, and more 
generally to a future revision of the Semantic Web recommendations.

This report is an appendix to report HPL-2008-59. It gives a worked 
example of the construction used in the proof from that report. For 
finiteness, a reduced datatype map consisting of only xsd:boolean is 
used. Each of the graphs in the construction is listed explicitly,
with some redundancy eliminated. The final Herbrand graph contains about 
15,000 triples.

Received on Wednesday, 21 May 2008 10:12:29 UTC

This archive was generated by hypermail 2.3.1 : Tuesday, 6 January 2015 21:42:04 UTC