W3C home > Mailing lists > Public > public-rif-comments@w3.org > September 2009

Re: RIF-PRD and ECA-rules

From: Paul Fodor <pfodor@cs.stonybrook.edu>
Date: Tue, 1 Sep 2009 10:56:26 -0400
Message-ID: <10f1a0490909010756p1cb8e74bue9ffcfd26c9980eb@mail.gmail.com>
To: public-rif-comments@w3.org
Hi Chris,
Thank you for the response. You are right that FLD is the base for new
dialects.

> In Core, terms cannot be defined as equal in the language as you observe,
but you can implement rules that check the equality of existing terms, e.g
from some predefined dataset (maybe RDF).

I see. I realize that there are enough reasons why equality is allowed only
in the bodies of rules and facts (it is very expensive to be implemented for
deduction and many applications don't need this feature, and I'm aware of
very few systems that implement full equality, while the most popular rule
systems don't handle equality at all).

Thank you, regards,
Paul.

On Fri, Aug 28, 2009 at 9:32 AM, Chris Welty <cawelty@gmail.com> wrote:

> Hi Paul,
>
> Thanks for the feedback. In general we do not have specific plans for new
> dialects as a working group, but new dialect definitions are sought based on
> FLD.
>
> In Core, terms cannot be defined as equal in the language as you observe,
> but you can implement rules that check the equality of existing terms, e.g
> from some predefined dataset (maybe RDF). E.g. imagine a rule conditioned on
> a person's membership level in some organization.
>
> -The RIF WG
>
> > From: Paul Fodor <pfodor@cs.stonybrook.edu>
>
> > Dear RIF list members,
> >
> > I'm interested in the RIF standard and I have a few comments (mostly on
> the
> > RIF Production Rule Dialect document). Overall, I like the RIF hierarchy
> as
> > a method for exchanging rule-like information between different systems
> and
> > it, definitely, looks as a complete and a mature material. I think that a
> > very strong point for RIF is its condition for safeness of rules, so that
> > the logical semantics of RIF-BLD and the operational fixed-point
> semantics
> > of RIF-PRD coincide.
> >
> > My first suggestion is to use the RIF-PRD (with its forward chaining
> > operational semantics) as the base dialect for a reaction rules dialect
> > which also incorporates events in the style of Event-Condition-Action
> (ECA)
> > rules: on <event> if <condition> then <action> (of course, if there is
> > enough interest from the community for this need). It seems as a logical
> > extension for the production rules dialect (PR are ECA rules without the
> > event part). Another special case of this ECA dialect would be triggers
> > (i.e., ECA rules without the condition part). Moreover, the event part of
> > this dialect could also address Complex Event Processing (CEP) for rules
> for
> > the event composition (which can be also interchanged rules under RIF).
> >
> > My second comment/question is about equality. The core dialect only
> allows
> > equality in rule premises which is puzzling to me because I don't see how
> is
> > then used (how terms are 'defined' as equal).
>
Received on Tuesday, 1 September 2009 15:15:04 GMT

This archive was generated by hypermail 2.2.0+W3C-0.50 : Tuesday, 1 September 2009 15:15:06 GMT