Re: RDF semantics: problems with entailment rules rdfs5 and rdfs9

>Two rdfs entailment rules do not seem to be stated
>in sufficient generality.
>This is indicated by the following two counterexamples
>to the rdfs entailment lemma.
>
>Example: graph G consists of triples
>   x subPropertyOf y
>   y subPropertyOf z
>where x y and z are blank nodes.
>In this case the semantics shows that
>G rdfs-entails the triple
>   x subPropertyOf z
>However this triple is not derived by rule rdfs5,
>since this rule requires x y and z to be URIs.
>
>
>A similar problem occurs with rule rdfs9.
>
>Example: graph H consists of triples
>   x subClassOf y
>   z type x
>where z is a blank node.
>According to the semantics, H rdfs-entails the
>triple
>   z type y
>This triple is however not derived by rule rdfs9
>since this rule requires z to be a URI.
>
>
>Herman ter Horst

Further to the above, I have amplified the textual explanation of the 
conventions, and gone through section 7 to bring the conventions in 
line in all the rule tables, so that the use of uuu, xxx, etc. is 
uniform throughout and corresponds exactly and unambiguously to the 
minimal syntactic needed to state the rules.

This was the original intention, but details got buried in multiple edits.

BTW this has the amusing consequence that 'www' is now used as a 
schematic for any subject, which has a kind of poetic justice to it.

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@ihmc.us       http://www.ihmc.us/users/phayes

Received on Friday, 7 November 2003 12:19:43 UTC