RDF Datatyping Stake Burning on 2002-05-17
- Literal nodes are syntactically and semantically untidy.
- Literal nodes denote things in the universe, but their meaning
is contextual (literals do not have globally consistent meaning)
- Literal nodes denote a single value in the context of an associated
datatype.
- The only new vocabulary term is rdfs:Datatype.
- The nature of an instance of rdfs:Datatype is as defined in
the former stake-in-the-ground.
- There are only two idioms provided:
- the inline idiom (global/implicit)
- the datatype property idiom (local/explicit)
- The semantics of the datatype property idiom remains unchanged
from the former stake-in-the ground.
- rdfs:range is used to associate a datatype with an inline literal.
- A property is presumed to have at least one, possibly undefined,
datatype range.
- The following closure rule is provided by the datatyping MT:
?s ?p ?o .
?p rdfs:range ?d .
?d rdf:type rdfs:Datatype .
?o rdf:type rdfs:Literal .
implies
?s ?p ?x .
?x ?d ?o .
Thus ...
The following entailment holds:
?d rdf:type rdfs:Datatype .
?p rdfs:range ?d .
?s1 ?p "LLL" .
?s2 ?p ?x .
?x ?d "LLL" .
implies
?s1 ?p ?v .
?s2 ?p ?v .
or in terms of Lbase, the n triple
<u1 s1 "FFF"> if there is no explicit data type, maps to
(exists dx s1(u1, df("FFF", s1,dx)))
and if it has an explicit data type X, maps to
s1(u1, df("FFF", s1, X))
where df is the data-typing function that maps a literal in context to its decontextual meaning.
The following entailment holds:
?s1 ?p "LLL" .
?s2 ?p "LLL" .
implies
?s1 ?p ?v .
?s2 ?p ?v .
The following entailment does not hold:
?s1 ?p1 "LLL" .
?s2 ?p2 "LLL" .
implies
?s1 ?p1 ?v .
?s2 ?p2 ?v .