Re: RIF vs Rule Language

On Dec 9, 2005, at 11:52 AM, Pascal Hitzler wrote:

> Hassan,
>
> I very much apprechiate your point - the term "formal semantics" is 
> easily interpreted in too narrow a sense. But let me just add (and 
> emphasize, as I assume that you agree with me on this) that procedural 
> semantics alone is not satisfactory for a semantic web context, as it 
> usually binds a language to a specific engine, and is thus not in the 
> spirit of interoperability and interchange.

That's a bit strong, isn't it?  I'm not sure if you meant there to be a 
nuance with "procedural" vs. "operational" semantics, but in 
programming theory, an operational semantics works by describing the 
behavior of an abstract machine at varying levels of granuality, 
sometimes at quite a high level. So, there is a sense that this "binds" 
the language to a specific (abstract) engine, but you can show 
equivalence of various abstract machines (at some level of 
granularity). Maybe to bring it back to the more familiar, you can show 
the equivalence of two proof theories (e.g., that they validate the 
same theorems and lets assume a deduction theorem) without detouring 
through a semantics.

(Or to make it simpler...and operational semantics that was useless for 
optimization, e.g., for showing that substituting a faster function for 
a more expensive one in a certain context produced the same actual 
results, would be a damn poor semantics indeed :))

I don't say it's *appealing*, but it can be made to work. And sometimes 
it is appealing :)

Cheers,
Bijan.

Received on Friday, 9 December 2005 23:24:44 UTC