A constructive formal semantics for SPARQL

Attached is a paper I have written containing thoughts on the best
way to express the formal semantics of SPARQL.  In it I contrast
what I call a "destructive semantics" vs. a "constructive semantics".
My former attempt, posted in
http://lists.w3.org/Archives/Public/public-rdf-dawg/2006AprJun/0170.html
was a destructive semantics.  After further reflection, prodded by a
remark by Eric, I came up with the attached constructive semantics,
which I now believe is superior.  The attachment also incorporates
what I take to be consensus on the semantics of OPTIONAL and
that the proper scope for blank node identifiers in queries is
rule [21] FilteredBasicGraphPattern, rather than just the basic graph
pattern as currently in section 2.5.1 "General framework".  I hope that
this will be a step forward in specifying the semantics of SPARQL.
Please note that I do not believe there are any "bomb-thrower" issues
in this paper such as my much-criticized message on "existential vs.
concrete semantics"
( 
http://lists.w3.org/Archives/Public/public-rdf-dawg/2006JulSep/0008.html ).
This is just an attempt to express the consensus as best I can determine it.

Fred

Received on Friday, 28 July 2006 00:08:21 UTC