Re: [RIF] Extensible Design --> RIF semantics

Vincent, Paul D wrote:
> Francois - I interpret this as:
> - the rules being interchanged have semantics
> - the actual interchange mechanism itself does not.
>
> This is presumably a problem for those who identify RIF itself as a future semantic web rule language, rather than an interchange mechanism for whatever that language turns out to be. 
>
> Or am I missing something here?
>
>   

These are clear statements -- but they do not fully convince me. For
expressing the semantics of the rules being exchanged, we need a
formalism. I can only think of thwo such formalism:

A. Specifying the rule engine, eg Prolog 3, or one of the semantics
known from the literature, e.g. stratified semantics with structural
equality.

B. Rely on a more refined formalism for specifying the semantics.

A is fully system, product, language, etc. dependent. Furthermore, how
to convert a ruleset after, say Prolog 3, in a rule set after a PR
language, is in almost all practical cases fully unknown. Therefore, my
conclusion is that A looks nice -- but leads to nothinhg usefull.

B, in my opinion, means that the interchange language comes along with a
formalism for expressing semantics. This amount to be a language with a
well-defined semantics (possibly allowing variations).

Thus, my understanding is that a usefull interchange mechanism requires
a rule language with semantics.

Francois

Received on Friday, 5 May 2006 09:34:00 UTC