[Scl] Re: Report on Common Logic

[andersen at ontologyworks.com]

Hi John..

What follows may sound like I'm angry at you but let me say in advance that
this is not the case.  Just want to make sure my view on this is clearly
understood.

> I wouldn't say that it's "too complicated", but the current
> SCL document is not easy to read, especially by non-
> mathematicians (who are our major customers).  If Chris
> can find a way to make the SCL document substantially
> more palatable, that would be good.

I don't feel very qualified to comment on technical issues right now, at
least until I get caught back up with SCL.  However, I think worrying about
"customer" acceptance is symptomatic of overly-academic thinking.

The original reason behind the pursuit of CL was to provide for the "rest
of us" (those more in the KIF than in the DL camp) to have a firm
foundation for a practical ontology expression language.  I can assure you
that 99% of those on the DL bandwagon don't understand much if anything
about the formal underpinnings of it.  Why do we expect more from CL
"customers"?  As I've said, there is at least one *commercial* user of a
KIF-like language with seqvars, and that's Ontology Works.  I would include
Cycorp and others, but I mention my company in particular (what I
personally am interested in may differ) because we have only commercial and
no academic interests.

What we are interested in is a document to which we can point and say "look
- we're following a standard!" so all the folks to whom that's important
can feel better.  We also want that document to support the features that
we find useful in *commercial* work, and seqvars are one of those features.

Bottom line is that the complexity of the spec for seqvars doesn't bother
me at all.  I will take the time to understand it because as an implementer
it is important for me to get it right.  But for the bulk of CL (as it is
for the DL) "customers" it won't matter a bit.  Why should they care about
the complexity of a semantic specification that they will never need to
read?

One footnote - I don't think any potential acceptance problem for CL or its
instance languages versus DL is due to a complexity issue in the semantics.
It's due to the fact that DLs resemble both frame systems and OOP systems. 
Users of these systems are far more numerous than those comfortable writing
FOL axioms and they find DLs comfortable, whatever their other virtues are
for ontology authoring.  I don't have an answer for how to win such a
popularity contest but I don't want to do it at the expense of tossing a
feature that's so near and dear to my heart.  

  .bill


--------------------------------------------------------------------
mail2web - Check your email from the web at
http://mail2web.com/ .