[SCL] Re: scl:comment [was: Re: SCL spec question]

[Pat Hayes]

>SCL-ers - {Pat},
>
>For context, I asked Pat about identity conditions on formulas named 
>inside (scl:comment ...) forms.
>
>Pat..
>
>See below...
>
>On Sep 8, 2004, at 1113, Pat Hayes wrote:
>
>>I was hoping we could use strict syntactic identity. In fact, for 
>>some purposes, even token identity. So for example this would make 
>>sense:
>>
>>In BillsOntology:
>>
>>.....(forall (?x)(p ?x))
>>
>>In PatsOntology:
>>
>>.... (scl:comment "this is the same as the axiom in Bill's 
>>ontology" (forall (?x) (p ?x)))
>>
>>BUt they are still different axioms as far as the syntax is concerned.
>>
>>BTW, this question of what constitutes identity is an open issue 
>>for text with markup, as in XML, as far as I know. Som=eone needs 
>>to look at this issue carefully, its on my stack somewhere....
>
>Yeah, it's a tough one.  My natural tendency is to say it's an 
>ontological question and try to bring some of that machinery to 
>bear, but I think that's overkill for a standard like this and might 
>well slow down progress towards ISO status.
>
>I think a reasonable compromise might be a notion of identity that 
>works within an SCL document.  It seems a reasonable assumption that 
>if I say:
>
>(scl:comment "bill likes this" (forall (?x) (p ?x)))
>(scl:comment "pat doesn't like this" (forall (?x) (p ?x)))
>
>in the same document, they pretty clearly refer to the same axiom. 
>This does make for a problematic situation for collaborative 
>development where I enter an axiom (with a comment) one day, and you 
>on the next write another comment (intended to be) for the same 
>axiom.

Well,  the commenting syntax is designed recursively to allow 
'nesting' of comments. Thus on three successive days the axiom might 
look successively like:

(forall (?x) (p ?x))
(scl:comment "bill likes this" (forall (?x) (p ?x)))
(scl:comment "pat thinks bill is crazy" (scl:comment "bill likes 
this" (forall (?x) (p ?x))))


>
>Another (I think reasonable) compromise, would be to introduce 
>non-denoting constants to name axioms.

Well, wait a minute. If they name axioms don't they denote them?

>  I say non-denoting because I'm trying to avoid quantification over formulas.

Hmm. I see why you want to avoid it, but these 'names' would have to 
be distinguishable from names that really do denote, right? How?

>  It has the additional advantage of not having to retype the axiom. 
>So I say something like:
>
>(scl:formulaID a1 (forall (?x) (p ?x)))
>
>This gives users (or collaborators in the development) of an 
>ontology unambiguous access to the axiom for the purposes of 
>documentation.  Uses of such constants within a document need to be 
>unique.  With this, you could say in the same file, e.g.:
>
>(scl:comment "bill likes this" a1)
>
>Or if Pat includes the "bill" ontology containing a1:
>
>(scl:comment "pat disagrees" bill:a1)
>
>This is, in general, a useful mechanism for commenting on any 
>constant introduced.  I don't think it's a good idea to embed this 
>in the term syntax, since there seems to be no principled place to 
>put the comment.

? Seems to me that one can put a comment anywhere.

>  What are the ramifications of doing it this way instead:
>
>(scl:comment "biological fatherhood relation" bill:fatherOf)
>
>One big difference could be that bill:fatherOf denotes, but for 
>something as mundane and useful as making comments I think there's 
>good cause to special case the function of scl:comment

Hmm, maybe. I basically wanted to allow comments (annotations more 
generally) but make them logically transparent/invisible. You are 
suggesting a convention which treats them in a particular way. Since 
they don't denote, the role of axiom labels seems to be orthogonal to 
any logical role they might have. So this is a convention that could 
be added transparently to SCL without modifying it, I think.

One issue is that if comments are transparent then your example

(scl:comment "biological fatherhood relation" bill:fatherOf)

is logically just a bare constant name, which currently isn't a legal phrase.

Need to think about this some more. Would you say that this example 
should be treated like an atomic sentence?

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 ihmc.us       http://www.ihmc.us/users/phayes