W3C home > Mailing lists > Public > public-rdf-dawg@w3.org > October to December 2009

Re: [TF-ENT] URIs for entailment regimes in service descriptions

From: Birte Glimm <birte.glimm@comlab.ox.ac.uk>
Date: Sun, 1 Nov 2009 17:19:19 +0000
Message-ID: <492f2b0b0911010919i47c10687i63fd2a3b609a4b80@mail.gmail.com>
To: Sandro Hawke <sandro@w3.org>
Cc: SPARQL Working Group <public-rdf-dawg@w3.org>
Sandro,
one of the problems with just saying OWL is that reasoners that use
Direct semantics only work on a subset of RDF documents (those that
fall into the OWL DL syntactic fragment). Most will not even parse
non-OWL DL documents and OWL EL reasoners might not even parse OWL DL
documents.
Since these URLs are for the service descriptions, it is not the user
that specifies this. Any system that says it does RDFS entailment will
of course do RDF and simple entailment. If on the other hand a user
queries a system that says it does simple entailment, then you would
not get any answers that depend on RDF or RDFS entailment.
For Simple, RDF, RDFS, OWL RL, OWL RDF-Based, that all works quite
nicely because these regimes all take any RDF graph as input, but OWL
Direct semantics is only defined for a subset of RDF graphs, those
that correspond to OWL 2 DL ontologies.
Thus, OWL 2 DL reasoner will probably reject many non-OWL 2 DL
documents and OWL 2 EL and QL reasoners even further restrict the set
of supported ontologies. It is not so easy to just fall back to simple
or RDF(S) entailment in this case. One could not use URLs for the
profiles, but than an OWL EL/QL reasoner cannot communicate that it
only works on a subset of OWL 2 DL.
Birte

2009/11/1 Sandro Hawke <sandro@w3.org>:
>
>> > http://www.w3.org/ns/entailment/OWL-Direct
>> > http://www.w3.org/ns/entailment/OWL-RDF-Based
>
> I'm concerned about the market fragmentation here, and wondering if it's
> possible to just say "OWL".  Probably not.  But we should do something
> to make sure that where the entailments are the same, it doesn't matter
> which semantics the user specifies.  In particular, I'd like users to be
> able to say one thing (eg "OWL") and get (possibly incomplete) reasoning
> using all the regimes Ivan listed.  There's also the question of getting
> RIF reasoning here, which Axel and I have been talking about.
>
> If we're not careful, I expect people will think that if they implement
> only one of these entailment regimes, they should simply ignore queries
> stating one of the others.  If fact, for many queries and graphs, they
> have also implemented the other.  If they don't recongnize that fact,
> the functionality to clients will be much reduced.  Put differently,
> clients should ask for the entailment level they want, confident that
> they'll get something sound, but not necessarily complete, with respect
> to that regime.  (And perhaps with completeness flagged, somehow.)
>
> (Sorry if I'm jumping in not really understand the history here.)
>
>       -- Sandro
>
>
>> In a sense ../entailment/OWL-Direct is the same as ../owl2-profile/DL
>> and ../entailment/OWL-RDF-Based is the same as ../owl2-profile/Full.
>> That is why DL and Full are not really profiles, but an OWL EL
>> reasoner for example also applies the direct semantics of course. To
>> be consistent with OWL though, not creating the Full and DL URIs might
>> make sense.
>
> I suggest that "OWL-Direct" and "OWL" are the right terms here, since
> the RDF Based Semantics are a superset of the OWL-Direct semantics.
>
>
>> Birte
>>
>> 2009/11/1 Ivan Herman <ivan@w3.org>:
>> > Birte,
>> >
>> > I was not at the call, sorry about that.
>> >
>> > What I try to propose to the SW Coordination Group is the following set
>> > of URI-s
>> >
>> > http://www.w3.org/ns/entailment/Simple
>> > http://www.w3.org/ns/entailment/RDF
>> > http://www.w3.org/ns/entailment/RDFS
>> > http://www.w3.org/ns/entailment/D
>> > http://www.w3.org/ns/entailment/D
>> >
>> > http://www.w3.org/ns/owl2-profile/Full
>> > http://www.w3.org/ns/owl2-profile/DL
>> > http://www.w3.org/ns/owl2-profile/EL
>> > http://www.w3.org/ns/owl2-profile/QL
>> > http://www.w3.org/ns/owl2-profile/RL
>> >
>> > This seems to cover what the RIF group requires, it seems to be o.k. for
>> > =C2=A0OWL (although you might want to convince Ian that the profile ones =
>> are
>> > necessary) and seems to cover what SPARQL needs.
>> >
>> > What may be controversial whether the Full and DL profiles are really
>> > necessary; after all, DL is a syntactic subset of OWL ie, it is a
>> > profile although the term is not used. But if we do not need it, we can
>> > just not create those for now.
>> >
>> > Does this work for you, ie, for SPARQL?
>> >
>> > Thanks!
>> >
>> > Ivan
>> >
>> >
>> >
>> >
>> > Birte Glimm wrote:
>> >> Hi all,
>> >> in the telecon we shortly discussed whether we would need two URIs for
>> >> the two OWL Semantics (RDF-Based and Direct (<- Model Theoretic,
>> >> Description Logics)) or URIs for each OWL Profile.
>> >> For RDF-Based seantics, we have two profiles:
>> >> OWL Full and OWL RL
>> >> OWL Full handles all legal RDF graphs and so does an OWL RL system.
>> >> For OWL RL, if the input belongs to a certain fragment (the OWL RL
>> >> fragment), then the system is guaranteed to sound and complete,
>> >> otherwise the system might be incomplete.
>> >>
>> >> For Direct semantics we have three profiles:
>> >> OWL QL, OWL EL, and OWL DL
>> >> Any OWL DL reasoner can handle all three (but not arbitrary OWL Full).
>> >> OWL EL reasoners can also handle OWL QL, and OWL QL reasoners do only
>> >> OWL QL.
>> >> Inputs outside of the supported fragment will be rejected, i.e., an
>> >> OWL QL system works only on inpus that fall into the OWL QL fragment.
>> >> OWL EL systems will accept inputs that fall into the EL (and thus also
>> >> into the QL) fragment, etc
>> >>
>> >> The problem with using just one URI per semantics is that OWL QL and
>> >> EL systems will possibly reject many input ontologies that are OWL DL
>> >> because they are outside of their fragment. If we have just one URL,
>> >> then I cannot know what the system will accept. It is trial and error.
>> >> For RDF-Based semantics it seems to be less of an issue, but for
>> >> Direct Semantics it would make more sense IMO to have different URIs
>> >> and then it would be a bit wired to have only one for RDF-Based
>> >> semantics, but three for Direct Semantics.
>> >>
>> >> Birte
>> >>
>> >
>> > --
>> >
>> > Ivan Herman, W3C Semantic Web Activity Lead
>> > Home: http://www.w3.org/People/Ivan/
>> > mobile: +31-641044153
>> > PGP Key: http://www.ivan-herman.net/pgpkey.html
>> > FOAF: http://www.ivan-herman.net/foaf.rdf
>> >
>>
>>
>>
>> --=20
>> Dr. Birte Glimm, Room 306
>> Computing Laboratory
>> Parks Road
>> Oxford
>> OX1 3QD
>> United Kingdom
>> +44 (0)1865 283529
>>
>



-- 
Dr. Birte Glimm, Room 306
Computing Laboratory
Parks Road
Oxford
OX1 3QD
United Kingdom
+44 (0)1865 283529
Received on Sunday, 1 November 2009 17:19:59 GMT

This archive was generated by hypermail 2.3.1 : Tuesday, 26 March 2013 16:15:40 GMT