Some ground laying

Perhaps I don't really have a sense of what the pace is supposed to be, 
but if we plan to have anything substantial by ISWC, we need to get the 
ball rolling. Plus, upcoming telecon.

****Some preliminaries****

First off, my default position is that relatively little *needs* to be 
said, if anything, from a technical point of view. I don't think the 
current RDF specs, for example, are perniciously deficient on some 
general "meaning" front. I think there it's quite easy to say some 
wrong things, things which are generally harmful or simply 
embarrassing. I'd like to avoid that.

Second, I'll be using Tim's initial raising of the issue as my primary 
foil, perhaps with some other bits from past interactions with him. I 
note that http://www.w3.org/2001/tag/ilist#rdfURIMeaning-39 cut out a 
lot of stuff that roots some of the problems I see, and avoids issues I 
think Tim would acutally want on the table. So, for this moment:
	http://lists.w3.org/Archives/Public/www-tag/2003Jul/0022.html
will be my main text.

Third, there seems to be a whole slew of possible issues in this 
issuing raising. And for each identifiable issue/subissue, I think 
there are a number of questions we might seek to answer:
	What is the issue?
	Does it concern implementers, users, software agents, or human authors 
or readers?
	Is (currently) a matter for specificaiton? Or best practices advice? 
Or just left to the community?
	Is there "an" answer, or are there a set of choices with different 
trade offs?
	Is there some group who might properly claim jurisdiction (e.g., RDF 
Core)?

****Some issue framings****

**There is a specter hauting the Semantic Web...the specter of 
meaninglessness**

This may be an issue cluster.

It has been claimed (ok, I weasel) that without "something more" (some 
thing easy to add, in fact) RDF documents will be meaningless (or only 
have formal meaning which is the next worst thing to meaningless, 
apparently).

* So, are RDF documents, as currrently specified, in any pernicious 
sense meaningless?
* Are RDF documents, if meaningless in whichever sense, also useless? 
Are the interopt issues?
* If the meaning isn't specified by the specs (thus "spec" meaningless) 
are there huge problems with their meaning being application dependent?
* Is the meaning "almost there" (i.e., just needing a little bit of 
"glue" spec) or is the gap wider than that?

--------------------------------------
**The role of the Predicate**

Is a predicate the prime determiner of the meaning of an RDF assertion, 
aka triple?

(I say, no. I see no reason to prefer a relation oriented view over a 
Frame/subject oriented view.)

(This seems deeply critical, but sorta vague, to Tim's cluster of 
views, witness this false statement from the issue raising:

"""The OWL specification is a vocabulary of properties allowing an RDF 
document to say things about RDF Properties""")

If you use a URI in a predicate position in an RDF document, are you 
committed to whatever the URI owner says that predicate (if it is a 
predicate) is? Or to the ontology she gives for it? What is the nature 
of that commitment?

Note that these last bits generalize to the use of URI period.

Note that I think this is a matter for the working groups. Indeed, 
WebOnt already has taking *some* steps to defining which uses of a URI 
imply committment to a defining ontology (i.e., owl:imports) and has as 
an issue to do more.

There are some aspects of this that probably belong to RDF core or a 
similar group, to wit, RDF Core lets us know when two triples must be 
*co* asserted (they are in the same graph, and neither is encoded, 
e.g., in a datatype value, then if one is asserted so too is the other) 
but they don't tell us what it is for the graph to be asserted. In 
section 4 of Concepts and Abstract Syntax, there was some talk about 
"publishing", e.g., on a website, but that's pretty vague to me.

Ok, so this is an issue cluster. Predicate dominance, uri use,  
ontological commitment, and assertion seem to be the 4 issues.

--------------------------------------
**The URI-id-one-thing thing**

I imagine I don't have to say too much about this. :) It is an issue, 
it's likely a TAG issue.

------------------------------------
**The role of natural language (and other things)**

 From the raising:

"""This information, directly or indirectly acquired, may be
human-readable and/or machine readable, the latter including for
example ontological statements in OWL, or rules, or other logical
expressions."""

 From the WWW BOF, I got the very strong impression that the content of 
rdfs:comments were part of the constitutive definition of a term (when 
placed in the right document) regardless of the nature of that content. 
As an example of practice, I was pointed to:
	http://www.w3.org/2000/10/swap/log

It would be interesting to determine what exactly this schema can and 
does establish about the meaning of those terms.

----------------------------------

There's a lot in:
"""3) There may be some need to clarify frequent misunderstandings by
making some things clear.""""

Or rather under it, but I don't have the current energy to disintangle 
them at the moment.

Here's a start, anyway.

Cheers,
Bijan Parsia.

Received on Friday, 19 September 2003 12:06:00 UTC