[Top] [All Lists]

Re: [ontolog-forum] Webby objects

To: ontolog-forum@xxxxxxxxxxxxxxxx
From: John F Sowa <sowa@xxxxxxxxxxx>
Date: Mon, 19 Nov 2012 08:32:24 -0500
Message-id: <50AA34E8.20107@xxxxxxxxxxx>
On 11/18/2012 7:06 PM, Pat Hayes wrote:
> Ian's argument here is really an argument for *efficient*
> decidability, and (to give him credit) this is exactly
> what he is a world leader in providing.    (01)

I agree.    (02)

The only disagreement is over the issue of restricting the language
to make it impossible to write undecidable statements.    (03)

The solution they adopted in OWL DL is to force every model to be
expressible as a tree, because their proof of decidability only
works on tree structures.    (04)

That means you say that a benzene molecule has 6 carbon atoms, but
you can't say the atoms are connected in a ring -- because a ring
is not a tree.  It would get their theorem prover hung up in a loop.    (05)

They buy their beds from Mr. Procrustes.    (06)

> Ian's view is that logics that aren't decidable are simply faulty,
> broken, inadequate, not of interest, etc.. Any "reasonable" logic
> should be decidable.    (07)

Ian is very intelligent, but he needs to leave his ivory tower
and take a job where he writes application programs.  There are
many of ways of using logic, and theorem proving is *not* the
most important:    (08)

  1. Stating a query.  An FOL query against a relational DB is
     not only decidable, it can be answered in polynomial time.
     If the fields are indexed, the most common queries can be
     answered in logarithmic time.    (09)

  2. Stating a constraint.  The issues about queries also apply
     to constraint checking.    (010)

  3. Stating preconditions and postconditions for procedures.
     These are also reducible to constraints.  For many applications,
     a good compiler can derive an efficient procedure from them.    (011)

  4. Stating formal definitions for humans.  This is a task that
     the Z notation people have been doing for years -- some of
     them on very large, mission-critical systems.    (012)

  5. Knowledge representation.  Systems like Cyc can use and reuse
     the same information in many different ways -- including all
     of above and more.    (013)

> And as Ian had proposed the wording and I was objecting to it
> on what most people thought were arcane technical/logical/academic
> details, I lost the vote.    (014)

This is why I believe that the six-month review process for an ISO
standard is important.    (015)

With the ISO procedure, Bob MacGregor, Doug Lenat, Roger Schank,
and the entire world would have a chance to review the text
and make their objections known.    (016)

That would certainly have slowed down the process.  But for
something as important as a standard that is supposed to govern
the entire web for years, you can't accept a decision based
on the personal preferences of somebody who had never written
an application program in his life.    (017)

> I don't know what the complexity class of OWL Full is. Those who
> care about such things are usually working on description logics
> in any case.    (018)

Those people have been publishing their ideas in obscure journals that
nobody who has a day job ever reads.  And that's where they belong.    (019)

John    (020)

Message Archives: http://ontolog.cim3.net/forum/ontolog-forum/  
Config Subscr: http://ontolog.cim3.net/mailman/listinfo/ontolog-forum/  
Unsubscribe: mailto:ontolog-forum-leave@xxxxxxxxxxxxxxxx
Shared Files: http://ontolog.cim3.net/file/
Community Wiki: http://ontolog.cim3.net/wiki/ 
To join: http://ontolog.cim3.net/cgi-bin/wiki.pl?WikiHomePage#nid1J    (021)

<Prev in Thread] Current Thread [Next in Thread>