W3C home > Mailing lists > Public > public-rdf-wg@w3.org > November 2012

Re: Sloppy inference rules

From: Pat Hayes <phayes@ihmc.us>
Date: Wed, 7 Nov 2012 09:17:35 -0600
Cc: Richard Cyganiak <richard@cyganiak.de>, Guus Schreiber <guus.schreiber@vu.nl>, RDF WG <public-rdf-wg@w3.org>
Message-Id: <80C56D18-CFCC-49DE-8C45-3B970ECE3705@ihmc.us>
To: Ivan Herman <ivan@w3.org>

On Nov 7, 2012, at 7:29 AM, Ivan Herman wrote:

> 
> On Nov 7, 2012, at 06:20 , Richard Cyganiak wrote:
> 
>> On 30 Oct 2012, at 16:15, Pat Hayes wrote:
>>> So my question to the WG is, how disastrous would it be if the RDFS inference rules were stated in terms of a 'sloppy' version of RDF with the syntactic restrictions removed, i.e. with the following grammar for triples?
>>> 
>>> triple ::= term term term .
>>> term ::=  IRI | blanknode | literal
>> 
>> +1
> 
> Let us remind ourselves, by the way, that (per agreement at the F2F) the RDF(S) inference rules are the subject of a separate, non-normative Working Group Note; the normative semantics continues to be the model theoretic one. Ie, if, for the purpose of the notes, such generalized triples are used, that does not affect RDF normatively.

Indeed. It is still an ugly situation, but we can live with ugliness. 

> Having such 'generalized' triples in for the purpose of the inference rules is therefore perfectly o.k., but I would also like to see two things added to the note:
> 
> - drawing attention on the fact that these triples go beyond RDF triples, and they are there for the purpose of getting the rules working
> - making it clear that the rule engine is supposed to throw away non-RDF triples at the end of processing when returning an expanded graph.
> 
> I also realize that doing this means that, formally, we can forget about those confusing rules that 'replace' literals with blank nodes. Which is fine for the rule set but it may be worth keeping a note in the text making clear to implementers that, if they rely on RDF comformant environments, that is a possible approach they may use (actually, as of today, I would probably not even use blank nodes there, because that is always a scary thing to refer to, but rather skolem URI-s, now that we have them...)

Hmmm. This illustrates the danger of saying, as we now do (to keep things simple) that skolemizing "keeps the same meaning". In fact, if we are being logically strict, it doesn't. So, no, you can't just use skolem URIs instead of bnodes in inference rules. The resulting rules would not even be logically valid. 

BTW, I am completely at a loss to understand why anyone can find blank nodes "scary". They are about the simplest idea imaginable. Think of them as a mark made on a surface. 

Pat


> 
> Ivan
> 
>> 
>> Best,
>> Richard
> 
> 
> ----
> Ivan Herman, W3C Semantic Web Activity Lead
> Home: http://www.w3.org/People/Ivan/
> mobile: +31-641044153
> FOAF: http://www.ivan-herman.net/foaf.rdf
> 
> 
> 
> 
> 
> 
> 

------------------------------------------------------------
IHMC                                     (850)434 8903 or (650)494 3973   
40 South Alcaniz St.           (850)202 4416   office
Pensacola                            (850)202 4440   fax
FL 32502                              (850)291 0667   mobile
phayesAT-SIGNihmc.us       http://www.ihmc.us/users/phayes
Received on Wednesday, 7 November 2012 15:18:13 GMT

This archive was generated by hypermail 2.3.1 : Tuesday, 26 March 2013 16:25:52 GMT