summary of changes to semantics doc

Peter was right in noting that a vital condition had been left out of 
the RDFS semantics, this being that ICEXT(I(rdfs:Literal)) should be 
a subset of LV. This has now been fixed; also,  any recognised 
datatype must be a subclass of rdfs:Literal.

Typed literals with language tags are now handled by the semantic equations.

Rule rdfs11 has been renamed rdfD0c and moved to the datatype 
section. Also two new rules rdfD0a/b have been added to handle 
language tags in typed literals.

The rules rdf3a/b for handling XML literals have been rewritten to 
handle the bug noticed by Jan. (The result is kind of ugly, but it is 
correct and this way I didn't need to rewrite all the recursion 
lemmas; and there is a remark to the effect that this could be 
replaced by a canonicalization operation provided that it is applied 
to antecedents and consequents; similar comments have been added to 
the discussion of datatype normalization in section 4.3)

The Lbase axioms for datatyped literal handling have been largely 
rewritten and a few other of the axioms revised in the light of 
Graham's observation. The newer ones are a little longer but more 
obviously in correspondence to the model theory. The axioms now have 
an explicit predicate for describing language tags, as suggested by 
Peter.

Ive added the term 'token' to the glossary.

------

Pat


-- 
---------------------------------------------------------------------
IHMC					(850)434 8903 or (650)494 3973   home
40 South Alcaniz St.			(850)202 4416   office
Pensacola               			(850)202 4440   fax
FL 32501            				(850)291 0667    cell
phayes@ai.uwf.edu	          http://www.coginst.uwf.edu/~phayes
s.pam@ai.uwf.edu   for spam

Received on Monday, 20 January 2003 00:11:28 UTC