On Feb 1, 2010, at 3:53 PM, John Madden wrote:
Hi Oliver,
I don't want to speak for Eric, and I'm not even sure I've
accurately represented his point here. Nor am I sure that they *are*
different scenarios.
Instead of saying they are or aren't, let me throw out a scenario
that concerns me, and in the context of which they *might* be
different, and ask for people to comment.
Suppose I publish a medical document in English, and purport to
attach to it an "official" RDF/OWL graph.
A problem (sidebar):
Is this possiible? If so, what makes it "official"? Not sure. Maybe
the graph is connected to the document
with grddl transform that I have published, which makes explicit
the rules I use to go from English to RDF.
Maybe the graph contains some assertions that reference some
provenance/trust vocabulary, such that
triples are contained in it that assert -- in some way -- that it
is a privileged graph. Maybe the RDF/XML
document has an XMLSig on it and I've stated to the world that I
nobody is allowed, under penalty of my
wrath, to represent this document by RDF/OWL graphs except provided
that ones I've signed are provably
consistent with them (which is an inane thing for me to do, and not
only because I don't know anybody who
is afraid of my wrath.)
(For a medical document, it might not be *me* that insists on this
claim; it might be my employer/hospital.
They don't want people attributing meanings to the document other
than those they have had a chance
to approve, because they don't want somebody claiming the RDF/OWL
they published led to a subsequent
adverse event (by, e.g. being used in a decision support system at
some later time that attributed a different
meaning to some vocabulary item). So for example they might only
allow locally defined classes properties
to be used in the graph.)
Anyway, I somehow put a stake in the ground and loudly assert that
**this** is what it means.
Now later Eric comes along and wants access to my document, and many
others in a bunch of different databases, because he is aggregating
data about patient encounters for a research project of his. In his
SPARQL query, he does a bunch of joins, that in effect are mappings
between the local vocabularies in the existing RDF graphs and
whatever vocabularies he's attempting to aggregate to. In this
process, my "official" RDF triples get left behind.
Is this a possible scenario? Where does it fail? Is it that the
SemWeb doesn't support any notion of an "official" graph? Is it that
there is no such thing as an "official graph" at all (on the sem web
or anywhere else)?
It doesn't, and there isn't. The SWeb position on official is exactly
the same as the Web position, which might be summed up in the phrase
Caveat Lector. Publication is easy and free and unfettered, and
requires no imprimateur or legitimacy. So, its up the reader of what
is published, to decide whether or not to accept it, whether or not it
is trustworthy, etc.. This is why the top level of TimBL's layer cake
is labelled "trust" rather than, say, "authority" or "legitimacy". And
this is, of course, much like publication elsewhere, at least in the
free world.
Now, however, that said, the Web (and hence the Sweb) does give some
useful structure, in that every URI has an owner, and its not a bad
presumption that what the owner of it says using it might have more
legitimacy than what others say using it. in cases of doubt. (This
doesnt always work, but its not a bad heuristic.) So it might be a
good idea to take Eric's publications using Eric's URI as having more
authority on what that URI refers to, or at any rate being a more
accurate reflection of what Eric intended them to mean.
One of the motivations for the named graph proposal was to provide
exactly the kind of authoritative warrant of assertion that is
discussed in the above sidebar, by the way. [1]. Y'all might want to
check it out, I think it was quite ingenious. BUt no doubt too
complicated for immediate adoption by the Sweb community in its
current incarnation.
Pat Hayes
[1] www4.wiwiss.fu-berlin.de/bizer/SWTSGuide/carroll-ISWC2004.pdf
John
On Feb 1, 2010, at 3:14 PM, Oliver Ruebenacker wrote:
Hello,
On Mon, Feb 1, 2010 at 1:30 PM, John Madden <john.mad...@duke.edu>
wrote:
We had an interesting call in TERM today. One of the topics I
would like to boil down to the question "When does a document
acquire its semantics?" or, "when does a document come to mean
something?"
I argued the (admittedly intentionally) radical view that
documents have no semantics whatsoever until a reader performs an
act of interpretation upon the document, which in the Semantic Web
world would be the same as attributing an RDF/OWL graph to the
document.
Even if the author of the document attributes a a particular RDF/
OWL graph to her won document, I argued that this graph is not
privileged in any way. That others could justifiably argue that
the author's own RDF/OWL graph is incomplete, or flawed, or
irrelevant, or even incorrect. And the same is true of any
subsequent interpreters (i.e. authors of RDF/OWL graphs that
purport to represent the "meaning" of the same document).
Eric argued a really interesting point. He argued (and Eric,
correct me if I'm interpreting you wrong here), that semantics
instead come into existence (or perhaps *can* come into existence)
at the point when somebody executes a SPARQL query on a set of RDF/
OWL graphs. That is to say, maybe I'm wrong and semantics doesn't
even come into existence when somebody attributes an RDF/XML graph
to a document; but rather it only comes into existence when
somebody queries across (possibly) many graphs of many different
people.
What do you think?
Can you give an example were this difference is relevant?
Take care
Oliver
--
Oliver Ruebenacker, Computational Cell Biologist
Systems Biology Linker at Virtual Cell (http://vcell.org/sybil)
Turning Knowledge Data into Models
Center for Cell Analysis and Modeling
http://www.oliver.curiousworld.org
------------------------------------------------------------
IHMC (850)434 8903 or (650)494 3973
40 South Alcaniz St. (850)202 4416 office
Pensacola (850)202 4440 fax
FL 32502 (850)291 0667 mobile
phayesAT-SIGNihmc.us http://www.ihmc.us/users/phayes