Re: LC comment: sameAs in OWL QL

Ian,

you are right in saying that this comment is related to issue #133 and
it is my fault not to have checked that. It is also my fault that, at
the time, I did not pay enough attention. Apologies for both.

As I can see, we have three possible choices related to the exact shape
of QL:

1. Have QL include sameAs
2. Have QL include functional properties
3. Have QL include none of the above (letting implementation _add_ one
of the two)

#1 and #2 are mutually exclusive. The resolution on that issue (leading
to #3 which is the current status) was based on a compromise, because we
could not decide on either #1 or #2.

However, I think there are two issues that may lead us to reconsider this.

1. We have got a comment questioning our choice of profiles (Issue #26,
by Eli Lilly[1]). This was echoed by UvA's comment[2]. In a more general
sense, we have got several comments that require a better justification
for our design choices in terms of user requirements and communities. My
feeling is that the compromise solution we reached in resolving Issue
#133 is not an adequate one because it leads to unavoidable
interoperability issues (let alone the fact that the reasons are not
documented or at least I did not find them)

2. The resolution was taken in June 2008. Already at that time the
Linking Open Data community and project[3] had a significant push, but
this has only intensified since then, with, for example, a separate
conference being organized around this subject (let alone the various
workshops) or special issue planned for IJSW. Billions of RDF triples
are published with millions of, well, owl:sameAs statements used to
cross-link various datasets. My feeling is that the two 'small' profiles
(QL and RL) may both be attractive for that community if we carefully
consider their requirements. That could be a major push in bringing the
OWL and RDF worlds closer.

Ie, I would still like the WG to (re-)discuss this in view of those LC
comments...

Cheers

Ivan

[1]
http://lists.w3.org/Archives/Public/public-owl-comments/2009Jan/0033.html
[2]
http://lists.w3.org/Archives/Public/public-owl-comments/2009Jan/0042.html
[3]
http://esw.w3.org/topic/SweoIG/TaskForces/CommunityProjects/LinkingOpenData


Ian Horrocks wrote:
> 
> As Mike says, we already had an *extensive* discussion about this before
> deciding on the current solution. I don't see that the LC comment adds
> any new information, so why would be reopen the issue?
> 
> Ian
> 
> 
> On 4 Feb 2009, at 15:16, Mike Smith wrote:
> 
>>
>> On Tue, Feb 3, 2009 at 05:40, Ivan Herman <ivan@w3.org> wrote:
>>
>>> In the current version of OWL QL, owl:sameAs (if you prefer, same
>>> individual assertion:-) is disallowed. Having checked with Boris (and
>>> also chatted with Bijan) it seems that the reason is that this leaves it
>>> open to possibly extend QL *either* to include owl:sameAs *or*
>>> (exclusive 'or') functional properties.
>>>
>>> In my view, it would be better to include one or the other to the
>>> document to make the QL standard profile clearer and cleaner for users.
>>> Introducing a loophole of extra extension would reduce the usability of
>>> QL in my view, mainly in terms of interoperability.
>>>
>>> The LOD movement, for better or worse, has already made an extensive use
>>> of owl:sameAs in linking billions of triples stemming from public
>>> databases. In view of that use case, my proposal is to _add_ the same
>>> individual assertion into OWL QL. That would make OWL QL way more
>>> attractive for an important user community.
>>
>> This was one side of the discussion in [ISSUE-133].  I believe that
>> the email discussion linked to that issue (including the compromise
>> resolution [1]) are relevant when considering this LC comment.
>>
>> -- 
>> Mike Smith
>>
>> Clark & Parsia
>>
>> [ISSUE-133] http://www.w3.org/2007/OWL/tracker/issues/133
>> [1] http://lists.w3.org/Archives/Public/public-owl-wg/2008Sep/0017.html
>>
> 
> 

-- 

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