ontolog-forum
[Top] [All Lists]

Re: [ontolog-forum] Visual Notation for OWL Ontologies (VOWL)

To: ontolog-forum@xxxxxxxxxxxxxxxx
From: John F Sowa <sowa@xxxxxxxxxxx>
Date: Wed, 14 May 2014 17:37:00 -0400
Message-id: <5373E1FC.9060601@xxxxxxxxxxx>
William, Kingsley, and Ed,    (01)

JFS
> you need some way to represent plurals, such as "10 tentacles".    (02)

WF
> this is done ***very simply*** in extensions of FOL....
> I thought defining 'there exists exactly n' was a common exercise
> for students...  then say that all these xi are all pairwise unequal    (03)

Yes.  That is the simplest way to define 'exactly 2'.  But the
complexity of 'pairwise unequal' increases as N-squared.  The size
of a collection increases linearly with N.  Collections are simpler
for people to imagine and more efficient for computers to process.    (04)

> Five cats ate six fish may have an ***ambiguity*** problem, but
> once it has been resolved, it is easy.  If people are today
> publishing how to represent it???    (05)

I just mentioned one brief example that raises a host of issues.
For any specific case, it is easy to state a formula in logic
that represents it.  But analyzing the sentence, recognizing
possible ambiguities, representing each option in a general way,
and defining general methods for reasoning about them is far
from trivial.    (06)

> EG and CG and OWL ... HIDE the free variables.    (07)

EGs and CGs do not hide the free variables in their basic forms.
In English, a phrase such as '10 tentacles', has three implicit
components:  the collection of tentacles, each tentacle in the
collection, and the number 10.    (08)

An English speaker can refer to any of those components by terms
such as 'they', 'each', and 'how many'.  The CG "syntactic sugar"
that supports {*}@10 allows anaphoric references to all three.    (09)

KI
> Basically, how does this huge existing body of knowledge become a Web,
> once exploitable via public (World Wide Web) and/or private networks
> (intranets [LANs] and extranets [identity constrained WANs] ) ?    (010)

Most of the knowledge is in NLs.  The connections are in a variety
of computational forms.  The critical issue is to relate NLs to and
from any and every computable form.  My recommendation is to use
a version of logic that has mappings to and from all notations:
NLs for input, CNLs for output to human users, any computable
notation anybody has ever implemented.  For examples, see
slides 7 to 15 of http://www.jfsowa.com/talks/goal7.pdf    (011)

JFS
>> But you learned [anaphora] from your mommy long before that.
>> The Attempto examples may have focused your attention on it.    (012)

EJB
> No, I didn't.  My childhood taught me English.    (013)

For nearly everybody, the first and most important teacher is mommy.
People use the term 'mother tongue' for a very good reason.    (014)

EJB
> from experience, I can tell you that trying to explain to the
> human what was wrong with what s/he wrote is not easy.    (015)

Absolutely!  A typical university graduate today has far less
knowledge of grammar and logic than university graduates from
the 13th to the early 20th centuries.    (016)

I blame Bertrand Russell for the downfall of logic education in the
20th C.  He tried to get universities to stop teaching Aristotle
and begin teaching symbolic logic.  He got half his wish.    (017)

EJB
> Why would you bother to 'control' the language...    (018)

I would *never* attempt to force typical users to learn a CNL.
What I am recommending is the use of CNLs as *output* from the
computer to the humans.  For *input*, strategy #3 implies    (019)

  1. Design the system to interpret whatever people say or write,
     translate it to some version of logic, and generate a CNL echo.    (020)

  2. Carry on a dialog with the humans until they agree that the echo
     (which may be anything from a phrase to a paragraph) expresses
     what they were trying to say.    (021)

This is basically the way people communicate.  It's very rare for
people to understand what other people are saying without a dialog.
Just look at any thread on Ontolog Forum.    (022)

John    (023)

_________________________________________________________________
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    (024)

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