RE: No Standard Semantic Web Pragmatics?

> From: Sandro Hawke
> Sent: Monday, June 14, 2004 10:52 PM
> 
> > > > Again, why not the usual situation in the rest of the web? The
> > > > contents of Semantic Web web pages are asserted, just like other
> > > > web pages. 
> 
> > I don't agree that the default illocutionary force of a web page is
> > truth functional assertion of its propositions. The number 
> of counter
> > examples must number in the millions. Here are a few:
> > 
> > Sunset - http://www.nevis.columbia.edu/~hires/pictures/sunset.jpg
> > true? or false? Neither, for there are no propositions.
> 
> The actually proposition content served from that address is more
> like: there exists an image which is 1152 pixels across and 770 pixels
> high, with a JPEG encoding with the following parameters, ....
> 
> There's probably more connection between the URI and the image than
> "there exists....", but I'm not as sure how to formalize it.
> Pragmatically, if you do a web get on that URI, you'll be told (using
> standard protocols) that a particular image exists -- maybe that
> if-then is all the identification relationship we need.

We could also argue that when I write the expression, "Sunset!", I am 
asserting the logically true proposition "This expression has 7 
characters". Or if I am on the beach with my wife and exclaim, 
"Sunset!", we could argue that I was asserting the proposition 
"The air around my mouth has been disturbed by waves  of 222 
hertz frequency". But we are not asserting these propositions, 
they are facts about the communication substrate contingent upon 
the utterance. This may be another example of what the philosopher 
Paul Grice calls natural meaning. He gives the examples, "Those 
clouds mean rain." or "Those red spots mean measles".
 
> > "<foaf:Person rdf:about="#Peter_Frederick_Patel-Schneider">
> >   <rdf:type rdf:resource="#PerfectBeing" />"
> > - http://www-db.research.bell-labs.com/user/pfps/
> > asserted? No, an example of what could happen if you were required
> > to believe everything said by the owner of a URI you used.
> 
> Yes, it is asserted *by the web page*. 

I cannot accept the idea that it is web pages that make assertions. 
I don't believe you would make this claim in any other similar 
situation. "Your honor, I did not promise to pay him - the *contract* 
did.  And, no, I did not assert my agreement - it was my *signature* 
that did that."  It does not work in any situation I can imagine. 
Martha Stewart might have said, "I did not gave the order to sell, 
it was the *telephone handset* in my brokers office." Or after a 
marriage, "I did not say 'I Do', it was the *sound waves* from my 
mouth that said that." Things don't make assertions, agents do. Web 
pages are a medium of communication, not the communicator.

> It is not asserted by Peter
> (as far as I can tell). 

Of course not, he has clearly communicated what he meant by it.

> His relationship with his page is more
> complex than to simply assert everything it says.  Connecting formal
> statements to real people may be doomed to failure 

What does it mean to connect a formal statement to a real person? 
I'll guess you mean people assert them - but your trying to avoid 
the necessary condition of agency in that verb. On that basis, I 
think Euclid, Newton, Russell, and Godel may be a few well known 
examples of people "connected" to formal statements. So I don't 
think it is doomed to failure. How about Pat Hayes and the RDF 
Semantics Document? or Peter and OWL Web Ontology Language 
Semantics and Abstract Syntax?

> -- who ever means
> precisely and exactly and literally much of anything they say? 

I think this is entirely too pessimistic.

> -- but
> connecting formal statements to artificial-constructs/agents/web-pages
> seems just fine to me.

You think web pages are better at formulating and asserting formal 
statements than say Pat or Peter? because the former are artificial 
and the later are human? I disagree, I think humans are the best at 
and very nearly the only agents capable of making formal statements. 
 
> From there we can build together networks of trust, where one page
> asserts another.  Perhaps http://example.com would assert every page
> starting with "http://example.com", making trust reasoning scale a lot
> better.   (...essentially allowing one to reason about large sites,
> instead of individual pages.)  I'm not sure whether it's fair to use
> owl:imports as the way for one page to assert another....
> 
> > "302.0 Homosexuality is a type of
> > Sexual Deviation" - http://www.wolfbane.com/icd/icd8h.htm.
> > asserted now? No, a historical record of what the World Health
> > Organization believed in 1965.
> 
> Again, yes it is!  

I don't believe that *proposition* is still being asserted.

> The ICD Rev 8 (1965) is being faithfully
> represented on that page (as far as I know), and both it and the page
> claim that homosexuality is a type of "Sexual Deviation".  I would not
> expect that Wolfbane Cybernetic Ltd (the domain's owner) holds that
> view.

I think a better analysis would be to say that Wolfbane asserts 
that there is a proposition on that page that at one time (1965) was 
a statement asserted by the World Health Organization. That 
is an entirely different assertion than that made by the WHO 
about homosexuality back in 1965. The former is a present statement 
about a historical statement and the later was a statement about 
homosexuality. But if the default illocutionary force of an owner 
posting a web page was to assert the propositions it contains, you 
would have to say that Wolfbane was making those statements by 
asserting those propositions now. And its not.

> So I can say "I don't believe http://www.wolfbane.com/icd/icd8h.htm",
> or "I find http://www.wolfbane.com/icd/icd8h.htm offensive", without
> saying much of anything about wolfbane.com. 

Of course, you can comment on anything you want to, including a portion 
of a complex statement like this. And, yes, in a case like this, where 
Wolfbane is quoting an historical document of another organization, it 
is of less importance than in other situations. But it is still best 
to recognize Wolfbane's intent as I described it above can you really say anything intelligent about it. You can't 
eliminate Wolfbane and still claim to understand that page. There is 
no other way. That is the nature of communication. 

> I don't know of an RDF
> vocabulary for saying those things in RDF, but it shouldn't be too
> hard to specify (give or take trying to specify the meaning of things
> like the Liar Paradox which you could then construct).  
> 
> Of course I'd probably just publish a page which said that, so you
> wouldn't have a formal tie back to ME saying it, but DanBri suggested
> some ways one might attempt that.  Aside from the problems of
> formalizing human expressions, achieving strong on-line
> non-repudiation in a world with insecure computers is a hard problem.
> And one we don't need to solve for the Semantic Web.
> 
> Meanwhile, one could also publish a page which asserted that the ICD
> Rev 8 (1965) asserted that homosexuality was a type of what it called
> "sexual deviation"; that is, one could use some sort of quoting, given
> the right RDF vocabulary for it.  But that's just syntactic sugar,
> since we can just put something on another page to quote it.
> 
>      -- sandro
>

Received on Tuesday, 15 June 2004 20:21:25 UTC