Re: The RDF Approach to Indicating Language-In-Use

>From: Sandro Hawke <sandro@w3.org>
>Subject: The RDF Approach to Indicating Language-In-Use
>Date: Wed, 29 Oct 2003 10:09:05 -0500
>
>>  One issue jumped out at me during the last meeting, illuminating
>>  perhaps the central difference between Bijan's view and Tim's view.
>>  It was when Jim Hendler asked how one knew if an RDF document was
>>  supposed to be understood as an OWL document.  Bijan said to use
>>  application-specific logic [1].
>
>I don't think that this is the real difference between the two views, but
>let's continue.
>
>[...]
>
>>  I think Bijan is suggesting that systems need to work this out on a
>>  case-by-case basis.  That doesn't scale or support ad-hoc
>>  interoperation like we want; it's just the fallback if we can't come
>>  up with anything better.  I think he is advocating falling back now
>>  out of concern that in trying to address this problem we'll end up
>>  creating possibly bigger ones, such as by mandating "strong
>>  ontological committment".
>
>I would instead label Bijan's approach as the ``Do what you can'' approach.
>Applications that only understand RDF would process documents that looked
>like RDF as RDF.  Applications that can understand OWL (DL) would process
>documents that look like OWL (DL) as OWL.  More sophisticated applications
>could use whatever information is available to make the determination,
>including any standards that emerge.  This appears to scale very well.

I agree, and this is exactly the intuition behind the idea of RDF 
semantic extension. As long as each new language is a semantic 
extension in this (very weak) sense, then it is *harmless* to allow 
more ignorant processors to do what they can, since any conclusions 
they reach will not actually interfere with any conclusions that a 
more knowledgeable processor could reach. Not in a direct, 
in-your-face semantic contradiction sense, anyway (They might stir up 
some irrelevant mud, causing a more knowledgeable processor to have 
to do some extra syntactic work. Life's like that.)

Applying this idea of 'more knowledgeable' beyond formal model 
theories seems useful also. Let me suggest again that as a basic 
principle we say that any formal entailments made by SW reasoners *do 
not change* any'extra' meanings that the notation may be understood 
to have for any more sophisticated applications. Those include human 
readers, by the way, as well as what one might call non-SW software 
agents. Turning this around, its the requirement that any notions of 
'meaning' associated with a URI must be such as to be preserved by 
formally valid SW reasoning.

I think this simple rule/principle carries a lot of useful baggage. 
For example it requires that any 'meaning' attached to a URI by 
virtue of what it can be used to do in an HTTP protocol should be 
preserved by performing OWL inferences on expressions using that URI. 
It requires importing to be transitive, for example. Or take Sandro's 
evil-OWL example:

>How does it handle the case of evil-OWL?  Or someone using dc:author
as a synonym for rdf:type?

Well, the principle requires that your usage is preserved by any 
valid RDF reasoner. So if you use dc:author in this way and someone 
else can infer FOO from BAZ by using what you say and an RDF valid 
reasoner, then you have in effect sanctioned the inference of BAZ 
from FOO, and that in turn requires that any meaning - not just the 
'formal' meaning - which anyone might want to attach to BAZ had 
better be preserved in FOO. If you do something daft like this, that 
forces you to bite off more than anyone can chew, since very few 
other people will agree with you that dc:author is a synonym for 
rdf:type. Now, if you actually *state* that it is, eg

rdf:type owl:sameAs dc:author .

and publish that, then you are just as crazy, but nobody is obliged 
to use your assertion to draw any conclusions, and if they do so then 
its their responsibility, since the crazy conclusions are indeed 
valid, but have been explicitly made from your crazy assumption. If 
reasoners keep track of how their conclusions are derived then the 
craziness can be located.

Pat

-- 
---------------------------------------------------------------------
IHMC	(850)434 8903 or (650)494 3973   home
40 South Alcaniz St.	(850)202 4416   office
Pensacola			(850)202 4440   fax
FL 32501			(850)291 0667    cell
phayes@ihmc.us       http://www.ihmc.us/users/phayes

Received on Thursday, 6 November 2003 12:17:49 UTC