Re: [RIF] Extensible Design

From: Francois Bry <bry@ifi.lmu.de>
Subject: Re: [RIF] Extensible Design
Date: Thu, 04 May 2006 14:58:26 +0200

> Peter F. Patel-Schneider wrote:
> > The RIF is a language for interchanging rules.
> >   
> [...]
> > SEMANTICS
> >
> > None.
> >   
> 
> What does mean "Semantics: none"?
> 
> François

Something like:

The RIF will have no model theory.  The RIF will have no proof theory.  In
sum, the RIF will just be concerned with rules as data - not as syntactic
constructs in a logic.

Compliance of a tool with the RIF will be defined as
1/ being able to accept (all) valid RIF documents;
2/ being able to emit (a reasonable number of) valid RIF documents;
   and
3/ accepting an emitted RIF document results in the same internal "state"
   that caused the emission. 
The last condition would be measured by identifying a number of RIF
documents that are faithfully processed (i.e., no information is lost) and 
demonstrating that the emission of a RIF document after accepting one of
these RIF documents into a clean system results in the document itself.


Peter F. Patel-Schneider