Re: bNodes as graph identifiers

On Tue, 04 Jun 2013 22:41:34 +0200, Sven Kunze <sven.kunze@s2007.tu-chemnitz.de> said:

    > "Despite the use of the word “name” in “named graph”, the graph
    > name does not formally denote the graph.  It is merely
    > syntactically paired with the graph."  What is this
    > syntactically pairing good for if not for identifying the graph
    > itself?

Because there is a common use that goes like this:

<http://tu-chemnitz.de/sven#i> {
    <http://tu-chemnitz.de/sven#i> foaf:name "Sven Kunze";
                                   foaf:mbox_sha1sim "abcdef";
          ... other facts about sven ...
    <http://loc.gov/id/book> dc:author <http://tu-chemnitz.de/sven#i>.
}

this is the partitioning use and the graph label is often whatever is
arbitrarily considered to be the primary subject of the statements in
the graph. Since people are presumably a disjoint set from graphs,
that URI cannot sensibly denote both.

This use is reasonable, and quite useful since it makes it easy, for
example, to build a web page out of all the relevant information
expressed in triples about a particular entity, just by pulling that
particular graph. But it is incompatible with also using the graph
labels to refer.

-w