Re: draft public comment for OWL last call from RIF

After the email exchange with Bijan I think it makes sense to break this into 
two messages, so I'm back to this one and then I'll send another based on 
DaveR's suggested revisions.  Only a few moments to object, as I need to send 
this out today:

The RIF WG's primary mission is interchange, and interoperability between RIF 
and OWL is both an internal goal as well as, we believe, a goal of the semantic 
web community in general.  We believe that the semantic web standards should 
settle on a common interpretation of XML Schema datatypes.

We have reviewed the OWL definitions for the XML schema datatypes and have found 
them acceptable for RIF with one exception:  we cannot work with the 
redefinition of xsd numeric datatypes with overlapping (non-disjoint) value 
spaces.  While we all agreed the idea of e.g. "1.0"^^xsd:double and 
"1"^^xsd:decimal being the same entity makes sense, RIF adds a set of builtin 
functions and predicates to its chosen xsd's and these are based on a wide 
implementation base that assume disjointness of xsd value spaces.  Breaking 
these implementations would negatively impact interchange and significantly 
raise the "barrier to entry".


-The RIF WG

Dave Reynolds wrote:
> Sandro Hawke wrote:
>>> I'm happy with this level of detail on the disjointness of numerics.
>>>
>>> However, at the F2F we did not adopt *all* the other OWL proposals. 
>>> In particular, I thought we had agreed to not add owl:rational (which 
>>> is "at risk" in the OWL drafts) and we weren't minded to adopt the 
>>> specialist subtypes of xsd:string viz xsd:normalizedString, 
>>> xsd:token, xsd:Name, xsd:NCName and xsd:NMTOKEN. Technically they are 
>>> in limbo in that we did not (as far as I can tell) resolve to reject 
>>> them nor resolve to accept them.
>>>
>>> How about:
>>>
>>> [[[
>>> We have reviewed the OWL choices for the XML schema datatypes and 
>>> have found them acceptable for RIF with one major exception, and some 
>>> minor ones.
>>>
>>> Our primary concern is that we do not see how we can work with the 
>>> redefinition of xsd numeric datatypes with overlapping (non-disjoint) 
>>> value spaces.  While we all agreed the idea of e.g. "1.0"^^xsd:double 
>>> and "1"^^xsd:decimal being the same entity makes sense, RIF adds a 
>>> set of builtin functions and predicates to its chosen xsd's and these 
>>> are based on a wide implementation base that assume disjointness of 
>>> xsd value spaces.  Breaking these implementations would negatively 
>>> impact interchange and significantly raise the "barrier to entry".
>>>
>>> Of lesser concern we do not see value for our user base in adopting 
>>> owl:rational but note that is already At Risk in the current OWL2 
>>> drafts. We also do not see value in requiring support for the string 
>>> subtypes xsd:normalizedString, xsd:token, xsd:Name, xsd:NCName and 
>>> xsd:NMTOKEN.
>>> ]]]
>>>
>>> The point being we don't really care if they keep owl:rational and 
>>> the string subtypes but it would preferable if they were not required 
>>> within at least the OWL2 RL profile.
>>
>> Maybe we can be more explicit about that?  Rather than just say what
>> we're doing, say that we request OWL drop these types from the
>> RL-profile?   Then OWL-WG can just say "yes".
> 
> Fine by me. The reason I didn't include it in my draft was that I took 
> it that this message is a place holder to be followed up by further 
> discussion on what to do about it. After all they could also drop those 
> types entirely and we'd be happy with that :-)
> 
>> On numeric disjointness, maybe also point them to 
>> http://lists.w3.org/Archives/Public/public-rif-wg/2009Jan/0017.html
>> ?    Or I'll do that later.
> 
> Again either fine by me, though I'd be inclined to keep this message to 
> being the minimal marker that we have an issue and then follow up with 
> the details separately. Jos, of course, has the action to do the follow 
> up and we probably want to avoid triggering parallel threads of 
> technical discussion.
> 
> Dave
> 
>>
>>       -- Sandro
>>
>>> Cheers,
>>> Dave
>>>
>>> Chris Welty wrote:
>>>>
>>>> [RIFWG - comments please.  This is the message I propose to send to 
>>>> the OWL public comments list from RIF.  Should I get more technical, 
>>>> or is this sufficient.]
>>>>
>>>> The RIF WG's primary mission is interchange, and interoperability 
>>>> between RIF and OWL is both an internal goal as well as, we believe, 
>>>> a goal of the semantic web community in general.  We believe that 
>>>> the semantic web standards should settle on a common set of 
>>>> datatypes and a common interpretation of them.
>>>>
>>>> We have reviewed the OWL choices for the XML schema datatypes and 
>>>> have found them acceptable for RIF with one exception:  we cannot 
>>>> work with the redefinition of xsd numeric datatypes with overlapping 
>>>> (non-disjoint) value spaces.  While we all agreed the idea of e.g. 
>>>> "1.0"^^xsd:double and "1"^^xsd:decimal being the same entity makes 
>>>> sense, RIF adds a set of builtin functions and predicates to its 
>>>> chosen xsd's and these are based on a wide implementation base that 
>>>> assume disjointness of xsd value spaces.  Breaking these 
>>>> implementations would negatively impact interchange and 
>>>> significantly raise the "barrier to entry".
>>>>
>>>>
>>>> -The RIF WG
>>>>
>>>
>>
> 
> 
> 

-- 
Dr. Christopher A. Welty                    IBM Watson Research Center
+1.914.784.7055                             19 Skyline Dr.
cawelty@gmail.com                           Hawthorne, NY 10532
http://www.research.ibm.com/people/w/welty

Received on Friday, 23 January 2009 14:26:45 UTC