Re: Some comments on the spec of RDF Semantics

>Some comments on the spec of RDF Semantics:
>
>A. The semantic conditions and axiomatic triples for 
>rdfs-interpretation (section 3.3)
>
>1) IC should contains I(rdf:XMLLiteral).
>2) Add an axiomatic triple: rdf:XMLLiteral rdfs:subClassOf 
>rdfs:Literal. Because the following semantic conditions could not 
>guarantee that rdf:XMLLiteral is a subclass of rdfs:Literal:
>ICEXT(I(rdfs:Literal)) is a subset of LV
>ICEXT(I(rdf:XMLLiteral)) is the set of all canonical XML documents.
>2') Another way is to require that
>  ICEXT(I(rdf:XMLLiteral)) is a subset of ICEXT(I(rdfs:Literal)), 
>such as by specifying that
>ICEXT(I(rdfs:Literal)) is a subset of LV, and includes all canonical 
>XML documents and so on.

Your points are well taken. Other comments have suggested editorial 
changes related to these comments, so the exact form of the statement 
of the semantic conditions may be changed somewhat, but I will 
undertake to ensure that the final form of the equations does satisfy 
your comment.

>
>B. The rdfs-closure of an RDF graph
>The following is cited from section 4.2:
>[[The rdfs-closure of an RDF graph E is the graph got by adding 
>triples to E according to the RDF closure rules together with the 
>following rules:
>1. Add the RDFS axiomatic triples from the table in section 3.3, and 
>all the following triples. There are many other triples which are 
>true in every rdfs-interpretation, but they will be generated from 
>these by the closure rules.
>
>rdf:type rdfs:range rdfs:Class .
>
>rdfs:Resource rdf:type rdfs:Class .
>rdfs:Literal rdf:type rdfs:Class .
>rdf:Statement rdf:type rdfs:Class .
>rdf:nil rdf:type rdf:List .
>rdf:XMLLiteral rdf:type rdfs:Datatype .
>
>rdf:subject rdf:type rdf:Property .
>rdf:predicate rdf:type rdf:Property .
>rdf:object rdf:type rdf:Property .
>rdf:first rdf:type rdf:Property .
>rdf:rest rdf:type rdf:Property .
>
>2. ...
>
>3. ...
>
>]]
>
>The above eleven triples (except for: rdfs:Literal rdf:type 
>rdfs:Class .) can also be gained or generated from the RDFS 
>axiomatic triples in section 3.3 and the other closure rules.
>For example:
>The following are gained from the RDFS axiomatic triples in section 3.3:
>rdf:type rdfs:range rdfs:Class .
>rdf:XMLLiteral rdf:type rdfs:Datatype .

Yes, this is an editorial slip which others have noted. I plan to 
rewrite the prose of all the statements of the closure rules in a 
uniform way so that each set of rules automatically includes all 
previous ones, as well as the inference principle of existential 
generalization stated as an explicit inference rule.

>The triple "rdf:nil rdf:type rdf:List ." is from the RDF closure 
>rules (Point 1 of the section 4.1)
>
>The triple "rdf:subject rdf:type rdf:Property ." can be gained as follow:
>rdf:subject rdfs:domain rdf:Statement .(from the RDFS axiomatic 
>triples in section 3.3)
>rdfs:domain rdfs:domain rdf:Property .(from the RDFS axiomatic 
>triples in section 3.3)
>rdf:subject rdf:type rdf:Property . (the rule rdfs2)
>
>Similar to others.
>
>In addition,
>The triple "rdfs:Literal rdf:type rdfs:Class ." can also be 
>generated if we add an axiomatic triple: rdf:XMLLiteral 
>rdfs:subClassOf rdfs:Literal. as suggested in A.2).

A good point which I will check in detail in the final version.

>
>In summary,
>1)there is no need to explicitly add the above eleven triples,
>2)but need to add: rdf:value rdf:type rdf:Property .
>3)The point 2 of section 4.2 seems to be redundant, these triples are from
>the RDFS axiomatic triples in section 3.3.
>
>
>C. Typewriting error:
>1) rdfRV (should be rdfV) in the first paragraph.
>2) I(rdfs:comment) and I(rdfs:label) are repeated in the last row of 
>the first table in section 3.3, as follows:

Both now corrected.

>
>IP contains:
>I(rdf:type), I(rdfs:domain), I(rdfs:range), I(rdfs:subPropertyOf), 
>I(rdfs:subClassOf), I(rdfs:comment), I(rdfs:label), I(rdf:subject), 
>I(rdf:predicate), I(rdf:object), I(rdfs:member), I(rdf:_1), 
>I(rdf:_2), ... , I(rdf:first), I(rdf:rest), I(rdfs:seeAlso), 
>I(rdfs:isDefinedBy), I(rdfs:comment), I(rdfs:label), I(rdf:value)
>
>
>
>D. More like a question. Why not shift the rule rdfD 0c (section 
>4.3) to the category of the rdfs-closure rules(named as rdfs11)?

An excellent suggestion which I plan to adopt.

>
>The rule rdfD 0c means that every instance of rdfs:Datatype is a 
>subclass of rdfs:Literal (also section 2.4 in the spec of RDF 
>Schema).
>
>It seems reasonable to require the rdfs interpretation satisfying 
>this constraint.

Yes, indeed.

Pat Hayes
-- 
---------------------------------------------------------------------
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 Wednesday, 19 February 2003 12:17:05 UTC