"Semantics" vs. "No Semantics"

Dear All,

Thinking over the weekend of the "Semantics" vs. "No Semantics" issue
which arises again and again, I would like to submit a few simple views:

1. The RIF needs a specification of its meaning, e.g. how conjunctions
are expressed in RIF. No interchange language can be usfeull if its
meaning is not specified in some way.

2. This "specification of meaning" can be very abstract (= hight level),
and possibly will have to be so.

For me, and as it seems for m,any Computer Scientists, a "specification
of meaning" for a laqnguage is called "semantics". A language semantics
does not have to be specified in terms of, eg, models. And not all
models are "Tarskian models" (the salient feature of Tarskian models,
those of classical logic, is that they define the interpretation
funcxtions recursively on the formulas structures. Eg stable and
well-founded model do not follow this approach.)

I would suggest an agreement on the following:

- the RIF needs a semantics

- this semantics can be verty abstract and does not have to be definied
in terms of models or Tarskian models.

- the RIF probably does not need to specify a procedural semantics.

Is this fine with all?

François

Received on Monday, 8 May 2006 08:13:43 UTC