W3C home > Mailing lists > Public > www-rdf-comments@w3.org > July to September 2003

Re: Entailment rules development suggestion

From: pat hayes <phayes@ihmc.us>
Date: Tue, 5 Aug 2003 10:42:12 -0500
Message-Id: <p06001a07bb557eedff82@[10.0.1.3]>
To: Adrian Walker <adrianw@snet.net>
Cc: www-rdf-comments@w3.org

>Pat --
>
>There's a possibly naive suggestion, below, about an approach to the 
>entailment rules.
>
>Before I dive in the deep end with only weak dog paddle, do you know 
>please of anyone else who has tried this ?

Yes, and indeed I have done it myself. The Lbase translation in 
appendix A of the document is essentially this.

>Do you think it has any chance of success ?

Yes and no. In the long term this is the only rational way to 
proceed, in my view. In the short term however it introduces many 
complexities and new issues, if carried out in detail: for example, 
conventional logic makes no reference to datatypes; and RDF is much 
less expressive than full predicate logic, so an appropriately 
restricted notion of entailment is not easy to capture directly. 
Moreover, there was a felt need for a specifically RDF system of 
inference rules, which I have been attempting to satisfy. For 
example, a complete RDFS reasoner should be much easier to implement 
and more efficient than a complete full FOL reasoner running on a 
logical translation of RDFS (similarly for OWL, DAML, etc.) ; the 
Stanford group have had some success running a logical translation on 
the SNARK theorem-prover, however.

>
>Is it anyway too late for the W3C RDF Semantics document ?

Yes, I am afraid it is.

Pat

>TIA,   -- Adrian
>
>>Date: Sun, 03 Aug 2003 10:44:50 -0400
>>To: "Peter F. Patel-Schneider" <pfps@research.bell-labs.com>
>>From: Adrian Walker <adrianw@snet.net>
>>Subject: Re: dissatisfaction with the entailment rules development
>>Cc: www-rdf-comments@w3.org
>>
>>Peter --
>>
>>I have been following some of the discussion about RDF entailment, 
>>and I have a rather naive question please.
>>
>>Would it be helpful to do the following ?
>>
>>1) Map the RDF notation into ordinary predicate logic (or datalog, 
>>or other, as appropriate).
>>    (Perhaps using something like 
>>http://www.w3.org/TR/2003/WD-rdf-mt-20030123/#Lbase )
>>
>>2) Use the available large body of theory results, theorem provers, 
>>datalog processors etc to prove and run entailments.
>>
>>3)  Map back to RDF notation.
>>
>>4)  If direct processing of RDF entailments in RDF notation, 
>>without mapping in and out of logical notation is needed,  'compile 
>>down'  the steps 1, 2, and 3.
>>
>>
>>Hope this helps.                     -- Adrian
>>
>>
>>
>>
>>                                    INTERNET BUSINESS LOGIC
>>           Your English Business Rules Using Your Oracle Database
>>                                     www.reengineeringllc.com
>>
>>
>>
>>Adrian Walker
>>Reengineering LLC
>>PO Box 1412
>>Bristol
>>CT 06011-1412 USA
>>
>>Phone: USA 860 583 9677
>>Cell:    USA  860 830 2085
>>Fax:    USA  860 314 1029
>>
>>
>>>X-Originating-IP: [18.29.1.71]
>>>Resent-Date: Sun, 3 Aug 2003 09:29:52 -0400 (EDT)
>>>Date: Sun, 03 Aug 2003 09:07:24 -0400 (EDT)
>>>To: www-rdf-comments@w3.org
>>>From: "Peter F. Patel-Schneider" <pfps@research.bell-labs.com>
>>>X-Mailer: Mew version 2.2 on Emacs 21.1 / Mule 5.0 (SAKAKI)
>>>Subject: dissatisfaction with the entailment rules development
>>>X-Archived-At: 
>>>http://www.w3.org/mid/20030803.090724.07263263.pfps@research.bell-labs.com
>>>Resent-From: www-rdf-comments@w3.org
>>>X-Mailing-List: <www-rdf-comments@w3.org> archive/latest/3659
>>>X-Loop: www-rdf-comments@w3.org
>>>Sender: www-rdf-comments-request@w3.org
>>>Resent-Sender: www-rdf-comments-request@w3.org
>>>List-Id: <www-rdf-comments.w3.org>
>>>List-Help: <http://www.w3.org/Mail/>
>>>List-Unsubscribe: 
>>><mailto:www-rdf-comments-request@w3.org?subject=unsubscribe>
>>>
>>>
>>>
>>>I am deeply dissatisfied with the way the various entailment rules are
>>>specified in the RDF Semantics document (currently the version of 31 July).
>>>I had hoped that the entailment rules would finally end up as complete
>>>syntactic characterizations of entailment.  This would result in lemmas
>>>somewhat along the following lines:
>>>
>>>RDF(S) entailment lemma:  S rdf(s)-entails E if and only if there is a
>>>graph that can be derived from S plus the RDF (and RDFS) axiomatic triples
>>>by the appliation of the simple entailment rules and RDF entailment rules
>>>(and RDFS entailment rules) which is a supergraph of E.
>>>
>>>Instead the entailment lemmas are incomplete in a disturbing way.  The RDF
>>>entailment lemma defers to simple entailment, which makes it an incomplete
>>>characterization of rdf-entailment.  It would be much better to remove this
>>>incompleteness.
>>>
>>>The RDFS entailment lemma also depends on simple entailment, but also has a
>>>condition that S be rdfs-consistent.  This detracts considerably from the
>>>utility of the RDFS entailment rules.
>>>
>>>
>>>Peter F. Patel-Schneider
>>>Bell Labs Research
>>>Lucent Technologies


-- 
---------------------------------------------------------------------
IHMC	(850)434 8903 or (650)494 3973   home
40 South Alcaniz St.	(850)202 4416   office
Pensacola			(850)202 4440   fax
FL 32501			(850)291 0667    cell
phayes@ihmc.us       http://www.ihmc.us/users/phayes
Received on Tuesday, 5 August 2003 11:40:59 GMT

This archive was generated by hypermail 2.2.0+W3C-0.50 : Friday, 21 September 2012 14:16:32 GMT