W3C home > Mailing lists > Public > public-rdf-dawg@w3.org > January to March 2006

Re: Final text for Basic Graph Patterns

From: Pat Hayes <phayes@ihmc.us>
Date: Thu, 19 Jan 2006 16:20:14 -0600
Message-Id: <p06230909bff5b77bdd6f@[10.100.0.23]>
To: Enrico Franconi <franconi@inf.unibz.it>
Cc: RDF Data Access Working Group <public-rdf-dawg@w3.org>

>On 19 Jan 2006, at 18:33, Pat Hayes wrote:
>>Of course. You miss my point. I was simply following the general 
>>pattern of how SPARQL queries are defined, using our most recent 
>>attempt at an 'entailment-based' general form of definition, and 
>>applying that to OWL-DL as described in the OWL spec, and seeing 
>>what we finish up with. ( {:a rdf:type :b} is legal OWL-DL, under 
>>appropriate constraints, and is an instance of the query under 
>>binding of a variable to a legal OWL-DL identifier, so... ) I meant 
>>only that if one takes a 'natural' extension of SPARQL to OWL, 
>>keeping the basic form of the definitions but replacing simple 
>>entailment by OWL-DL entailment, then examples like this turn up.
>
>What I am also saying that you have to transport the syntactic 
>constraints that OWL-DL expressions have, to similar syntactic 
>constraints to SPARQL queries when using OWL-DL entailment. Namely, 
>in queries bnodes and variables are not in property position of any 
>triple , nor in object position of rdf:type triples,

Which is the key. OK, but this does not seem to me to be obvious, 
given your insistence that query variables should not be conflated 
with query bnodes. While it is (I will concede :-) DL-inappropriate 
to allow bnodes in rdf:type object position (actually we could argue 
this point, but that REALLY would be going off track on this thread), 
as that would be 'higher-order' (sigh...), surely it should be 
permissible to query against class URI names as well as individual 
names in an OWL KB, in any version of OWL? After all, one can do that 
much even in basic SPARQL; and there is no OWL requirement on closure 
or comprehension for class URI names, so the mere existence of all 
those 'anonymous' intersections and so on would be irrelevant. Which, 
BTW, would provide an even stronger and more direct case for the FUB 
position that bnodes should not be treated like query variables.

>and there is no rdf, rdfs, owl vocabulary symbol in the query with 
>the exception of rdf:type in property position.
>
>>The only reason for our discussing this matter at all *in this 
>>forum* is because some folk - notably, Bijan and yourself - have 
>>argued forcibly that the SPARQL definitions should be couched in a 
>>way which generalizes naturally to the OWL case, by replacing 
>>'simply entails' with 'OWL-DL entails' in the definitions. That is 
>>what I did.
>
>And in fact I have the same goal. But I also know the limits to put 
>in order for the machine to work.
>
>>>First of all, I have been sloppy in my statement, since OWL-DL 
>>>queries do not exist - we are defining them now here. Now, there 
>>>is *no* theoretical nor practical result that even considers 
>>>queries of that kind in the DL literature.
>>
>>I do not dispute this. As I said, I was simply applying the 
>>proposed SPARQL definitions to OWL-DL (as defined in the OWL 
>>specs). If this gets us to an area that has not yet been explored 
>>in the research literature, that seems to me to be in fact an 
>>extremely good reason for NOT including ANY mention of OWL in the 
>>SPARQL documents at all, other than perhaps an informative warning 
>>that this is an open problem area unsuitable for standardization at 
>>the present.
>
>In my text, I am proposing to have an informative statement saying 
>that a safe way to have a working SPARQL with OWL-DL entailment is 
>to restrict the scoping set B to include only URIs, and to have the 
>above syntactic restrictions to the SPARQL BGPs.

Good idea, provided only that we don't use the official label 
"OWL-DL" for this case which I think would be misleading. How about 
just calling it "simple OWL" or maybe "basic OWL" or some such 
qualification (?)

Pat

PS. We could call it SNOWY, since eliminating bnodes keeps us close 
to ground graphs, and snowy owls stay close to the ground and are 
very successful hunters and have few natural enemies. 
(http://www.owlpages.com/owls.php?genus=Bubo&species=scandiacus)

Or, on the other hand, perhaps not.

>
>cheers
>--e.


-- 
---------------------------------------------------------------------
IHMC		(850)434 8903 or (650)494 3973   home
40 South Alcaniz St.	(850)202 4416   office
Pensacola			(850)202 4440   fax
FL 32502			(850)291 0667    cell
phayesAT-SIGNihmc.us       http://www.ihmc.us/users/phayes
Received on Thursday, 19 January 2006 22:20:33 GMT

This archive was generated by hypermail 2.3.1 : Tuesday, 26 March 2013 16:15:25 GMT