Re: More on additional semantic information

"Jim Hendler" <hendler@cs.umd.edu> wrote:
>the problem is the points you make are true against some sort of 
>"general" ontological framework, but not against OWL where there is a 
>normative document, approved by the W3C as a web recommendation, that 
>outlines the specific semantics in question.  

I am not against any ontological framework.

>I'm also very confused 
>why you think that RDFS is somehow cleaner than OWL Full.  

I never even mentioned OWL Full, and I'm not claiming that.

>  However, please be clear what the "additional semantic information" 
>objective is about (as best I can tell) - it doesn't say create an 
>OWL query language, it says where there is an RDFS or OWL ontology 
>represent in the RDF graph that is being queried, there should be an 
>ability to expose some or all of those semantics 

yes, this is what I understand as well.

>-- a simple example 
>-- in a typical database if I query for
>  ?x a ex:animal
>I will only see those things that are exactly known to be animals -- 
>but if Bob is a bear (and bear is an animal), I wouldn't be 
>legitimized to answer "Bob" -- however, if I were querying an RDFS or 
>OWL backend with some inferencing, I would be very surprised if it 
>didn't include Bob.  

exactly, I agree. I wanted just to point out that if you have this 
requirement then the matter becomes suddenly not trivial at all. My simple 
example hopefully proved that.

My position is that I would be more than happy to have query language that 
answers correctly to cases like the above. But I also believe that this WG 
should concentrate (a) on laying down the semantic foundations for a 
general framework that includes such background additional semantic 
information, and (b) on proposing a concrete query language, effective 
evaluation procedures and specialised constructs only for the cases up to 
rdf(s). This is because only in the latter case (1) we may hope to have 
something properly working in the short/mid term, and (2) we comply to the 
charter that excludes any form of reasoning.

>I'd at least want a way to make it clear in the 
>protocol, if not in the query proper, as to whether I would want to 
>include inferred relations of some type.

Yes, no problem with that. But this presupposes that the query language we 
are setting up has a clear semantics wrt background ontologies. Having a 
clear semantics is must, it is the characterising feature of the work done 
in the semantic web. This is my above requirement (a).

>  Look, a lot of us spent several years designing OWL to sit correctly 
>on RDF, with the goal being, at least for some of us, that the 
>ontological properties created become first class citizens of the Web 
>-- for the DAWG to ignore this seems crazy to me.
>  So, while I would actually like "additional Semantic Information" to 
>be a requirement (not an objective), I can live with it as an 
>objective.  However, for it to be completely ignored would seem to me 
>we are not doing our job of supporting the entire RDF using 
>population, which includes us RDFS and OWL users.

I agree with you.

Cheers
--e.

Enrico Franconi                  - franconi@inf.unibz.it
Free University of Bozen-Bolzano - http://www.inf.unibz.it/~franconi/
Faculty of Computer Science      - Phone: (+39) 0471-016-120
I-39100 Bozen-Bolzano BZ, Italy  - Fax:   (+39) 0471-016-129

Received on Wednesday, 14 July 2004 00:58:44 UTC