Re: layering (was Re: Patel-Schneider Paradox ...)

>From: Frank van Harmelen <Frank.van.Harmelen@cs.vu.nl>
>Subject: Re: Patel-Schneider Paradox ...
>Date: Sat, 19 Jan 2002 17:22:40 +0100
>
>>  "Peter F. Patel-Schneider" wrote:
>>
>>  > Well, in some sense, no change is necessary.  RDFS does not appear to have
>>  > any problems in and of itself.  True extensions to RDFS can be 
>>made and also
>>  > do not appear to have any problems.  The problems only occur when a
>>  > same-syntax extension is mandated.
>>    ^^^^^^^^^^^^^^^^^^^^^
>>
>>  Indeed.
>>
>>  From my understanding of the discussions at Bell Labs,
>>  the problem does indeed only occur when OWL must be a "same-syntax
>>  extension", ie when OWL would be required to 
>>  - use RDF syntax, >*and*<
>>  - be a semantic >*extension*< of RDF
>>
>>  One of the proposals on the table was to keep the first and drop the second.
>>  OWL would then include most of the inferences sanctioned by RDF, but not
>>  all of them, making it not a strict extension. 
>>  (In particular OWL would loose some RDF-inferences that wouldn't make
>>  much sense from an OWL point of view anyway). 
>>
>>  This is the option that I would currently favour.
>>
>>  Frank.
>>     ----
>>  (Peter, please correct me if my understanding is wrong).
>
>Nothing wrong with your understanding here.
>
>However, such a relationship is dishonest, at least when it is described as
>an extension of RDF(S) that uses RDF syntax but violates the RDF(S) meaning
>for that syntax.

As long as it is honestly described, there is no dishonesty. Terms 
like 'extension' have many meanings and need to be defined carefully. 
However, there is certainly a way to honestly describe the 
relationship that Frank is suggesting, so this seems to be an issue 
over wording, not over content.

>  Further, in various discussions over the last week, see
>www-archive@w3.org for an archive of (most of) them, I've come to believe
>that such dishonesty is a very, very bad idea.

Dishonestly is a bad idea. But that's a moral observation, not a technical one.

>Making the relationship honest is trivial, at least technically.

Quite.

>It
>suffices to describe the syntax of OWL as N-triples instead of describing
>OWL as layered on top of RDF(S) but not using (all of) the semantics of
>RDF(S).  The non-technical issues are much harder to solve; I think that it
>would be necessary to have many very-noticable disclaimers that OWL is not
>an extension of RDF, or layered on top of RDF, or ....  The political
>issues may in fact be even harder to solve here---it may just not be
>possible to not say that OWL is layered on top of RDF.

Oh, crapola. The people who want to have the functional layering so 
much don't give a rat's appendage about the refinements of the 
semantic issues, and the people who do care about those issues can 
read the fine print. You are objecting to the obvious way out of this 
trap on the grounds that a badly-worded document might mislead 
someone. OK, then write a good document. This is a storm in a 
tea-cup; it just requires a good technical writer.

Pat


-- 
---------------------------------------------------------------------
IHMC					(850)434 8903   home
40 South Alcaniz St.			(850)202 4416   office
Pensacola,  FL 32501			(850)202 4440   fax
phayes@ai.uwf.edu 
http://www.coginst.uwf.edu/~phayes

Received on Friday, 15 February 2002 03:11:00 UTC