[SCL] Telecon on the 8th.

[pat hayes]

First, I have to apologize for not having ready the basic syntax 
document I promised to have ready at the last telecon; my time has 
been totally used up in travel and other things. Next week will also 
be almost entirely used up by a DARPA PI meeting, so I do not now 
expect to be able to make serious progress on writing until mid-April.

The document can be found, by the way, at

http://www.coginst.uwf.edu/~phayes/CL/SCL-guide.html

but it is in a very rough and incomplete form and will be changing 
without warning. Feel free to comment on any part of it at any time, 
of course, but please do not cite it or rely on it wihtout checking 
with me first. In particular the style currently adopted there for 
presenting the abstract syntax is abominable and will be changed.

------

In the meantime, however, let us try to continue with some momentum 
by having another telecon next Tuesday (8th)  at the same time, 0900 
CT/1000 EST/1600GMT/1700 CET, (US) 413-351-4140. Please say your name 
when you arrive as the phone does not make any auto-entry noises.

The last telecon meeting was sparsely attended but we had some useful 
discussions. One of the points raised by Tamel (Tamel please correct 
me if I have this wrong)  was that SCL is likely to be more 
acceptable to a larger audience if it is more conventional, and he 
would prefer therefore that the 'basic' language be as conventional a 
FOL as possible, and that 'extensions' such as row quantifiers, 
variadic relations and terms in relation position be treated as, 
well, extensions.

My own view on this is that it works for row variables but not for 
the 'higher-order' syntax, since the conventional syntax is so 
obviously a special case of the latter. I am working on a compromise 
presentation of the basic syntax which tries to avoid the 'boggle' 
effect of seeming too new to a reader familiar with the conventional 
logical notation.  (Apart from Tamel's worries, this issue has been a 
major stumbling-block for several people in the Webont community, as 
several of us know, so we need to pay it some attention.)

------

John, if you will be on the call, can you be ready to tell us how CGs 
might relate to this? That is, suppose that we define a subset of SCL 
which would most naturally map to and from CGs (a direct syntactic 
mapping, not merely a logical equivalence), what would it be like, 
exactly? I understand it would not have row variables in it. Would it 
allow a relation to have multiple arities? Allow functional terms? 
Allow variables in relation position in an atomic assertion? Etc.

-------

One issue I would like us to talk about this time is the 'special 
domains' issue raised in

http://philebus.tamu.edu/pipermail/scl/2003-March/000079.html

To see a (rather clumsy) example of this kind of thing in action, 
look at the Lbase use of 'special names' (XML literals, strings and 
integers)  at

http://www.w3.org/2002/06/lbase/20030116Snapshot.html#consistency

I think SCL needs a general-purpose way of allowing for such name 
domains to be added, and to provide a general syntactic framework for 
adding them.  Ideas, comments, etc, etc. all very welcome.

Pat

PS. For the record, here is a sketch of how a simple 'recursive' 
domain can be defined in 'full' SCL. One uses the row-variable 
induction principle, is the basic idea. God alone knows how to adapt 
this to something like well-formed canonicalized XML, however.

Char('a') and Char('b') and ..... and Char('z') and
String()
and
Char(?x) implies (String(@y) implies String(?x @y))

 From which it follows, for any sequence of characters c1,...., cn, 
that String(c1 .... cn), using the induction rule @I (where PHI[a b] 
is:  Char(a) implies String(b)  ).


-- 
---------------------------------------------------------------------
IHMC					(850)434 8903 or (650)494 3973   home
40 South Alcaniz St.			(850)202 4416   office
Pensacola               			(850)202 4440   fax
FL 32501            				(850)291 0667    cell
phayes at ai.uwf.edu	          http://www.coginst.uwf.edu/~phayes
s.pam at ai.uwf.edu   for spam