Re: What do the ontologists want

> > Did it never occur to you that natural language IS a
> > global information space? [...]
>
>Yes, but thankfully it passes :-)
>
>I think the distinction is that if you use natural language as your
>"global information space", then you have to somehow get a machine to
>grok it - you're trying to get machines to grok a huge body of
>infromation, and while the end results would be incredible, it's a big
>task.

Yes, of course. I am not suggesting using NL as a transfer protocol.

>So, why not start at the bottom and work your way up? That's
>what the Semantic Web, and RDF, affords us - a chance to create our
>own langauges, and tailor them to our implementations, rather than the
>KR approach, which has always been to use existing langauges and then
>see what comes out of it.

You really are talking through your arse, you know.  KR work has 
consisted largely of inventing languages and tailoring them to 
implementations, rather than using existing ones. The fact that 
almost all KR formalisms are equivalent to logical notations was 
discovered after the fact, sometimes rather painfully. You CAN start 
at the bottom, I guess, but why bother to, when people have already 
done some of your building for you?

>The KR folk have a great body of work behind them, and it's an
>intreguing field, and one that has contributed far more to the
>Semantic Web in general than any other sector, IMO. I agree with
>DanBri that the situation is not at all like WWW vs. hypertext people
>in '80, although there are a few resemblances, the main one of which
>(that Len Bullard often points out) is that sometimes it can be seen
>as reinventing the wheel. It must get frustrating to explain to people
>the essentials of logic ond ontologies to people who, try as they
>might, are simply not going to get it [welcome to the humiliating
>world of professional lecturing :-)]. But at the end of the day, there
>are advantages to be grounded in the Web - it's here, and it's now.

I have no idea what you are talking about. (Prolog is here, and now. 
KIF is here, and now. So?). The issue is not about being grounded in 
the web, it is whether the web is better off with RDF, or with 
something else.

>Less ambiguity.
>
> > Using RDF would be like trying to fill a reservoir using
> > teaspoons.
>
>That's what I mean - but we're not trying to fill a reservior here, we
>just want a glass of water.

Thats not what the professionals that I know seem to want. They 
actually have reservoirs of information already on the web, some of 
it using formal notations so complicated that would make your 
toenails curl, and they are starting to get impatient. Interestingly, 
I was talking yesterday to a guy who has started three successful 
companies in this area, who told me they had tried using XML and 
discovered that the notation was such a crock that about 90% of their 
transmission traffic was being used up sending meaningless notational 
strings back and forth, causing performance problems; so they just 
trashed XML and wrote their own notation. If you make a bad enough 
mousetrap, the world won't even take it as a gift.

Pat Hayes

---------------------------------------------------------------------
IHMC					(850)434 8903   home
40 South Alcaniz St.			(850)202 4416   office
Pensacola,  FL 32501			(850)202 4440   fax
phayes@ai.uwf.edu 
http://www.coginst.uwf.edu/~phayes

Received on Thursday, 17 May 2001 21:28:41 UTC