W3C home > Mailing lists > Public > public-rif-wg@w3.org > May 2010

Re: Urgent: Issue with RIF-Core EBNF Grammar?

From: Axel Polleres <axel.polleres@deri.org>
Date: Tue, 11 May 2010 18:49:27 +0100
Cc: "RIF" <public-rif-wg@w3.org>
Message-Id: <5164B163-9363-46A6-9570-1900FDDCB36C@deri.org>
To: Christian De Sainte Marie <csma@fr.ibm.com>

On 11 May 2010, at 18:27, Christian De Sainte Marie wrote:
> Axel, 
> 
> I do not think that changing Core at this point is a good idea (to say the least :-) 

I also don't think that this is a good situation, but it is still quite awkward.

> And, hopefully, we do not have to, anyway: if you need Core rules combined with class membership facts, you can import them (facts), e.g. from an OWL ontology. Does not that solve your issue? 

Good question... I wonder what that means, if the facts themselves are not expressible in RIF?

> </chair>As you now, I've always been of the opinion that facts (data and data models) are orthogonal to rules, and that their representations should, therefore, be kept separate; and, then, the way to use the rules is to combine them with facts<chair> 
> One reason why we did not want class membership formulas in Core rules heads is because allowing rules, in PRD, that could morph individuals without restriction (by asserting them as members of new classes) would have been problematic for some of the main PR engines.

The point is: if the membership facts don't "materialise" in RIF, how can any rule with a membership assertion in the body ever fire...?
(maybe that view is too operational)

> As a consequence, class membership assertions are allowed, in RIF PRD, only when a new individual is created: I do not know if we could have tried and designed a restriction to the same effect for the use of class membership formulas in Core rules heads. But Dave and the Core stakeholders were not very keen on class membership anyway, if I remember correctly. 
> 
> Cheers, 
> 
> Christian 
> 
> IBM
> 9 rue de Verdun
> 94253 - Gentilly cedex - FRANCE
> Tel. +33 1 49 08 35 00
> Fax +33 1 49 08 35 10
> 
> 
> 
> From:	Axel Polleres <axel.polleres@deri.org>
> To:	RIF <public-rif-wg@w3.org>
> Date:	11/05/2010 18:15
> Subject:	Urgent: Issue with RIF-Core EBNF Grammar?
> Sent by:	public-rif-wg-request@w3.org
> 
> 
> 
> 
> Hi folks, 
> 
> We are working on a parser with some students and I am afraid my student found something awkward in the RIF Core grammar, see mail below.
> 
> Indeed, I think he poked into a quite weird issue: 
> It doesn't make sense to allow class membership terms in rule bodies, if they can't appear at all in *any* facts.
> The current grammar and the restrictions in Section 2.3 though only allows uniterms and frames as facts.
> 
> 
> To repair this
> 
> 1) We'd need to change in Section 2.3 Formulas of RIF-Core:
> 
> * Equality terms and class membership terms cannot occur in rule conclusions -- they are allowed only in rule premises.
> -->
> * Equality terms cannot occur in rule conclusions -- they are allowed only in rule premises.
> * Class membership terms can only occur in rule premises or as ground facts.
> 
> 2) a proposal to fix the grammar in Section 2.6 would  be:
> 
> In the Rule Language grammar:
> 
>  CLAUSE         ::= Implies | ATOMIC
> -->
>  CLAUSE         ::= Implies | ATOMIC | GROUNDTERM '#' GROUNDTERM
> 
> 
> 
> sorry for spotting this now only, but I am afraid this is severe.
> the fix is not very problematic, though.
> 
> Axel                 
> 
> 
> 
> 
> Begin forwarded message:
> 
> > From: "Obermeier, Philipp" <philipp.obermeier@deri.org>
> > Date: 11 May 2010 16:26:50 GMT+01:00
> > To: "Polleres, Axel" <axel.polleres@deri.org>
> > Cc: "Marco Marano" <marcomarano83@gmail.com>
> > Subject: RIF-Core: EBNF, equality/memberhip facts
> > 
> > Hi Axel,
> > 
> > I found a minor error in the EBNF grammar [1] for RIF-Core (Altough,
> > this grammar is informative due to the lack of well-formedness checks,
> > it is also defined as strict superset of RIF-Core.).  Within this
> > grammar you cannot derive Equality or Membership (ground) facts since
> > the ATOMIC rule's rhs is restricted to atomic formulas excluding
> > Equality/Membership formulas. Apparently, this restriction is well
> > justified since ATOMIC may appear in rule heads (cf. IMPLIES rule's
> > rhs), for which Core forbids Equality and Membership formulas. In
> > conclusion, an introduction of a new ATOMIC_FACTS grammar rule extending
> > ATOMIC by Membership/Equality  would solve this issue w/o breaking the
> > restriction for atoms in rule heads.
> > 
> > Best
> > Philipp
> > 
> > [1]
> > http://www.w3.org/2005/rules/wiki/Core#EBNF_Grammar_for_the_Presentation_Syntax_of_RIF-Core
> > 
> > --
> > Philipp Obermeier
> > Digital Enterprise Research Institute, National University of Ireland, Galway
> > email: philipp.obermeier@deri.org
> 
> 
> 
> 
> 
> Sauf indication contraire ci-dessus:/ Unless stated otherwise above:
> Compagnie IBM France
> Siege Social : 17 avenue de l'Europe, 92275 Bois-Colombes Cedex
> RCS Nanterre 552 118 465
> Forme Sociale : S.A.S.
> Capital Social : 611.451.766,20 
> SIREN/SIRET : 552 118 465 03644
> 
Received on Tuesday, 11 May 2010 17:50:01 GMT

This archive was generated by hypermail 2.2.0+W3C-0.50 : Tuesday, 11 May 2010 17:50:02 GMT