Re: draft response for LC comment 31

O.k.

Ivan

Peter F. Patel-Schneider wrote:
> Thanks, Bijan, for reminding me that the discussion of Issue 111 did
> more than touch on this issue.  I've strengthened my wording in
> the response as it relates to the issue dicussion.
> 
> See http://www.w3.org/2007/OWL/wiki/LC_Responses/FH5 for the current
> version of the proposed response.
> 
> peter
> 
> 
> From: Bijan Parsia <bparsia@cs.manchester.ac.uk>
> Subject: Re: draft response for LC comment 31
> Date: Sat, 14 Feb 2009 14:14:54 +0000
> 
>> On 14 Feb 2009, at 11:40, Ivan Herman wrote:
>>
>>> Hi Bijan,
>>>
>>> the impression is that (1) the discussion in Issue 111 concentrated on
>>> the issue of signalling DL vs Full semantics
>> No. E.g.,
>>  <http://lists.w3.org/Archives/Public/public-owl-wg/2008Apr/0265.html>
>>
>>
>>> and that is where the
>>> sameAs^3 solution came up and (2) Frank's comment is more on the
>>> EL/QL/RL choices.
>>  <http://lists.w3.org/Archives/Public/public-owl-wg/2008May/0140.html>
>> """We might introduce owl:EL++, owl:DL-Lite, and owl:OWL-R-DL, but
>>> say that they are, from the point of view of the consequence > relation,
>>> equivalent to owl:DL. We should then say that we merged > the actually
>>> orthogonal concerns of selecting the consequence > relation and
>>> providing hints into one construct due to practical > reasons."""
>>> That was touched upon in the discussion in Issue 111
>>> but, as far as I can see and remember, the DL/Full issue took the upper
>>> hand.
>> DL/Full was important , but Profile signaling (i.e., in order to hint to
>> an editor to maintain that profile) was a key motivation and was
>> explored as well.
>>
>> The only difference with Frank's is that he doesn't want it to name the
>> *intended* profile, but the *actual* profile. But many, if not most, of
>> the same problems apply (consider if you import such assertions).
>>
>> Plus, the problem just isn't worth solving at this level. He's basically
>> asking that we define a format for a very specific sort of
>> (determinably) data about a document. Why not owl:numberOfClasses? This
>> is pretty useful information too.
>>
>> See again the TONES repository.
>>
>> There's no evidence that I can see that we need a special, in band,
>> standard place to record such metadata. I've done surveys of the OWL
>> Web, built and hosted species validators, etc.  and not run into
>> trouble. The danger of divergence between this metadata and the contents
>> of the document are real and could be significant. (see how mime type
>> sniffing is a critical part of web browsers).
>>
>> To put it another way, if we add this, I shall be recommending that
>> people ignore it and that it is bad practice to supply or rely on it.
>>
>>> Am I completely wrong?
>> I think so. The reason sameAs^3 was even discussed is because Sandro had
>> a very strong desire that OWL Full semantics intending documents be
>> syntactically detectable. Since the intendedProfile mechanism was deemed
>> to be inadequate for this (or any profile signaling for any reason) that
>> we were able to find a work around for his problem helped us move
>> on. But intendedProfile was, always, by me at least, also intended for
>> profile signalling.
>>
>>> (P.S. Shouldn't we enjoy or week end instead?:-)
>> Oh yeah :)
>>
>> Cheers,
>> Bijan.

-- 

Ivan Herman, W3C Semantic Web Activity Lead
Home: http://www.w3.org/People/Ivan/
mobile: +31-641044153
PGP Key: http://www.ivan-herman.net/pgpkey.html
FOAF: http://www.ivan-herman.net/foaf.rdf