[SCL] Approach to a concrete syntax

[pat hayes]

>Hi,
>
>Just read the intense discussion about the concrete syntax. I have 
>not been writing any
>modifications to the current drafts, new concrete syntaxes or 
>participated in the
>conf calls: wanted to look how things settle.
>
>It seems that the XML concrete syntax I wrote and Murray continued to work on
>has created a lot of contradictions, since people do have different 
>aims/viewpoints,
>roughly like this:
>
>- markup viewpoint
>- FOL viewpoint
>- RDF/RFS family viewpoint
>
>The current XML concrete syntax does not fit nicely into the 
>RDF/RDFS viewpoint
>and this has created uneasy feelings for people. In whatever way we 
>improve/modify
>the current XML concrete syntax, this problem won't change.
>
>It now seems to me that maybe we should drop the current XML concrete syntax
>(not that it should be REALLY dropped or that  Murray should drop it) but
>rather avoid putting any stress on it at all: not publish it in the 
>SCL papers once we
>create these.
>
>Instead we should perhaps really create the RDF/RDFS-friendly 
>version of concrete
>XML syntax instead and work on improving this syntax.
>
>Some time ago I sent a few very rough notes about RDF/RDFS-friendly-SCL-in-XML
>syntax. Thought more about it by now. We also have the real SCL draft by now
>(what we hadn't at the time of the previous SCL-in-XML draft).
>
>For example, the real SCL draft clearly specifies the RDF/RDFS-semantics style
>meaning of predicates: a predicate is converted to a constant in FOL and
>we use Pred or App constructors to transform SCL to FOL.

Wait; that is not my understanding of the intention. The pred/app 
translation is one way to render SCL into conventional FOL syntax, 
but it is not the primary or basic relationship between FOL and SCL. 
That should be the simple fact that FOL (by which I mean here 
conventional FOL syntax ) *is* a perfectly legal SCL (sub)language 
and inherits the SCL model theory unchanged without any need to 
provide a syntactic transformation.  There is no 'conversion' 
involved; predicates in SCL *are* constants.

>Since this approach
>seems to be more or less agreed upon (I was not sure about it before and
>asked a few times) it is consequently clear that the SCL is NOT meant to be
>a syntax for FOL in general

NO, it IS meant to be such a syntax. Also, I do not follow why you 
are calling this by the title RDF/RDFS, since RDF(S) does not require 
the use of the app/holds translation either.

>, but FOL usage in the context of RDF/RDFS.
>
>It seems that the latter viewpoint (RDF/RDFS prioritised) was assumed by
>the majority of the SCL group, but not by me (prioritising a FOL viewpoint)
>or Murray (prioritising a markup viewpoint). This difference is 
>counterproductive.
>I'll switch over to prioritising the RDF/RDFS needs.

Please don't; RDF(S) is marginal to the SCL project.

This seems to be an issue which we need a higher bandwidth to discuss 
adequately. Can you make a telecon next Tuesday at the usual time? 
(Or indeed any othe time, if you prefer: you choose.)

Pat
-- 
---------------------------------------------------------------------
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