> From: Xiaoshu Wang
>
> > From: David Booth
> > . . .
> > My overall comment: Yes! I believe a URI
resolution ontology could
> > significantly help address these problems, while
still
> > permitting URIs
> > to be based on the http scheme, thus facilitating
bootstrapping and
> > minimizing barriers to adoption.
> >
> I am very doubtful about the practicality of such an
ontology. First,
> consider the size. RDF is all about URI. If an RDF
document has n
> statement, there will be 3n URIs. If k statements are
needed to
> describe the resolution of one URI, the solution
> unnecessarily increased the KB k-fold. . . .
That would be prohibitive, but that is not what I meant. I
meant that k
statements would describe how to handle a large number of
URIs all
matching a particular http URI prefix. For example, if k
statements
indicated how URIs that begin with "http://xyzpurl.org?" can be
resolved, then any number of URIs beginning with that prefix
would be
syntactically recognizable as being resolvable using those
conventions.
So the knowledge base increase would only be an increment of
k, not a
factor of k.
This approach of using http URI prefixes is explained more
fully in
"Converting New URI Schemes or URN Sub-Schemes to
HTTP" at
http://dbooth.org/20
06/urn2http/ .
> Unless, perhaps, you can describe the composition of
URI - as
> a string.
Yes, that is my intent.
> But, RDF is used to describe resource, which is
identified a string.
> The meaning of these string should be specified
"outside" of the
> system.
It sounds like you are referring to the difference between
making a
statement about a URI as a string versus making a statement
about the
resource that is named by that URI, and I agree that
distinction is
important.
So to do the above, we will need to have ways to make
assertions about
URIs that match a particular pattern. In
h
ttp://dbooth.org/2006/urn2http/#multiple-ownersYou I
outlined how this
can be done:
[[
For example, if the term http://xyzco
nsortium.org/terms/xyzprefix is
defined to indicate that something is a specialized xyz HTTP
prefix,
then metadata served (indirectly) via http://foo.com?fum can
indicate
that "http://foo.com?" is a
http://xyzco
nsortium.org/terms/xyzprefix ,
and metadata served (indirectly) via http://bar.com?bee can also
indicate that "http://bar.com?" is a
http://xyzco
nsortium.org/terms/xyzprefix .
]]
However, I'm not sure what would be the best way to write
something like
this in OWL. Suggestions?
Incidentally, there was recently a proposal for a
"Protocol for Web
Description Resources (POWDER)" working group
http://www.w
3.org/2006/12/powder_charter (W3C member-only link)
that was circulated to the W3C membership for input, and one
of its aims
was to "develop a mechanism through which structured
metadata
('Description Resources') can be authenticated and applied
to groups of
Web resources". If this work is taken up by W3C, it
could provide
standard ways to express such assertions. In the meantime
we can define
our own ways.
David Booth
|