W3C home > Mailing lists > Public > w3c-rdfcore-wg@w3.org > April 2002

Re: The Cannes Entailment [was: Coming to grips with the entailment put forth by Jeremy]

From: Graham Klyne <Graham.Klyne@MIMEsweeper.com>
Date: Mon, 29 Apr 2002 16:21:16 +0100
Message-Id: <5.1.0.14.2.20020429153700.03bd6970@joy.songbird.com>
To: jos.deroo.jd@belgium.agfa.com
Cc: bwm@hplb.hpl.hp.com, patrick.stickler@nokia.com, w3c-rdfcore-wg@w3.org
Jos, I think you're right that it's a misunderstabnding...

I was responding to "We build a rule mechanism", which I understood to be a 
means to define rules *in RDF* (similar to log:implies in N3), and which I 
think would be a unacceptably substantial change to RDF at this stage.  If 
this means simply that we define some extra closure rules, then I withdraw 
that particular objection.


Thinking aloud... would a new closure rule work?

To get from:
    jenny age "10" .
    age rdfd:datatype xsd:integer .
to:
    jenny age _:x .
    _:x xsd:integer "10" .

We'd need a closure rule something like:

    { ?s ?p ?l . ?p rdfd:datatype ?d . ?l rdf:type rdfs:Literal . )
  log:implies
    { ?s ?p _:newx .  _:newx ?d ?l . }

which isn't quite what the datatyping proposal currently says, because the 
rule fires whenever the object of ?p is something denoted by a literal, not 
simply when an actual literal is used.

Is this the sort of thing you had in mind?  Hmmm... I just found your 
earlier posting:

[[
yes, that would be something like

   { :rule3a . ?p rdfd:dcrange ?d . ?p rdfd:valueProp ?q . ?s ?p ?o }
       log:implies
           { ?s ?q [ :f ( ?p ?d ?q ?s ?o ) ] } .

   { :rule3b . ?p rdfd:dcrange ?d . ?p rdfd:valueProp ?q . ?s ?p ?o }
       log:implies
           { [ :f ( ?p ?d ?q ?s ?o ) ] ?d ?o } .
]]

Is that valid N3?  What does { :rule3a . } generate?   ... cwm likes it, so 
I suppose it's OK;  doesn't generate any triples, so I suppose it's a 
device for labelling?  Hey, even my own N3 parser thinks it's OK.

I guess the [ :f ( ?p ?d ?q ?s ?o ) ] is some kind of analog of a Skolem 
function?  I guess you might need that for proof finding, but for deductive 
closure ... aha, I see -- I called mine "_:newx"!

And rdfd:valueProp relates the property-with-literal-range to a 
corresponding property-with-value-range?

Taking all that on board, I guess we're talking about pretty much the same 
idea here.

#g
--

At 02:55 PM 4/29/02 +0200, jos.deroo.jd@belgium.agfa.com wrote:

>[...]
>
> >>The Rule Fix
> >>============
> >>
> >>We build a rule mechanism, as in the know fix, into RDF.
> >
> >I think adding rules to RDF is a substantial change and out of
> >scope.  Especially this late in the WG cycle.
>
>Graham, this must be a misunderstanding
>We have already 10 RDFS-entailment and RDFS closures rules in
>http://www.coginst.uwf.edu/~phayes/w3-rdf-mt-draft-42402.html
>and at least 3 draft ones for the RDFD-MT
>that fix here is just maybe 2 extra proposed ones for the RDFD-MT
>
>--
>Jos

-------------------
Graham Klyne
<GK@NineByNine.org>
Received on Monday, 29 April 2002 11:42:04 EDT

This archive was generated by hypermail pre-2.1.9 : Wednesday, 3 September 2003 09:47:38 EDT