"However, in RDF, the thing identified by a URI with
fragment identifier does not bear any particular
relationship to the thing identified by the URI alone.
This differs from some readings of the URI
specification[6], so attention is recommended when
creating new RDF terms which use fragment identifiers."
I confess to have never liked this attempt to use fragment
identifiers in RDF to 'descend into meaning' rather than
to identify a structural fragment; I think it's inconsistent,
and leaves you no way to talk about structural fragments.
The notion that the resource identified by
"http://some.host/some.path#" and "http://some.host/some.path"
are completely unrelated seems pathological.
So not sure 'attention is recommended' captures
the necessary caution.
I think part of the problem is that the draft only
addresses URI references with fragment identifiers
when those URI references are used _as RDF terms_.
But what about other uses? If I have a web page with
<a href="http://some.host/some.path#concept">link
to a concept</a>
what might a legitimate response be to clicking
on that as a link? I can't tell from this document
or the documents it references.