Re: I can accept... (Was: Re: Objective 4.6: additional semantic knowledge)

PatH wrote:
>> On Tue, May 11, 2004 at 04:52:52PM -0700, Rob Shearer wrote:
>>
>>>  4.6 Additional semantic knowledge
>>>  It should be possible for knowledge encoded in other semantic 
languages,
>>>  such as RDFS, OWL, and SWRL to affect the results of queries about 
RDF
>>>  graphs.
>>
>> In the interests of making progress, I'm willing to accept Rob's
>> version, above, of my original proposal.  In the current UC&R doc,
>> my original is 4.6 and Rob's variant is 4.6a.
>>
>> So, I'd be willing to accept a modest reworking of 4.6a:
>>
>> Additional Semantic Information
>>
>> It should be possible for knowledged encode din other semantic
>> languages -- for example: RDFS, OWL, etc. -- to affect the results of
>> queries about RDF graphs.
>>
>> In fact, unless someone objects, I'd like to make that the language in
>> the document and the version which we vote on at some point.
>
> Sorry to rock the boat, but this requirement worries me. It is 
> possible to interpret it as saying that a query-responder has a 
> licence to use RDFS, OWL, etc., content to respond to any RDF query. 
> I think that would be a very bad idea, if the query is unable to 
> specify which of the various languages are intended to be used.
>
> Moreover, the term 'semantic language' is not defined, and according 

A Google search for 'define:semantic' is returning results and one
for 'define:language' as well, but indeed, it can't find anything
for 'define:semantic language' (which is not a proof of course :)).

> to some versions of what SWeb meaning boils down to, it could include 
> unformatted English inside rdf:comment strings. Already we have a 
> potential disagreement over whether SWRL counts as a semantic 
> language.
>
> I would be happier if we did not have any such requirement. It seems 
> to me to be orthogonal to the query language design, and a matter to 
> be handled by an RDFS or OWL inference engine rather than by a 
> querying protocol. Our charter explicitly says that the RDF graph 
> being queried may be 'virtual' , so it could be an RDFS or OWL 
> closure or the virtual graph obtainable from a given graph from a 
> particular SWRL rule set: this provides us with a very neat way to 
> avoid this issue and keep our attention focused on querying graphs 
> rather than drawing inferences which may or may not be valid 
> according to a variety of different semantic frameworks.
>
> Finally, the wording suggests that OWL and RDFS are 'other' than RDF, 
> whereas in fact they are semantic extensions of RDF.
>
> At the very least, it ought to be possible to do a 'bare RDF' query, 
> explicitly rejecting any reliance on any other non-RDF implication or 
> machinery, and the requirement, if we keep it, should make this clear.

fully agreed

> Suggestion (if we must keep this requirement):
>
> Additional Semantic Information
>
> It should be possible for a query to indicate that the answers should 
> take into account knowledge encoded in RDF semantic extensions, such 
> as RDFS, OWL, etc..

OK, I would think that this is a requirement...


-- 
Jos De Roo, AGFA http://www.agfa.com/w3c/jdroo/

Received on Monday, 17 May 2004 16:15:01 UTC