Re: ISSUE-16 (entity annotations): A proposal for closing this issue without action

On 8 Jul 2008, at 16:00, Bijan Parsia wrote:

> On 7 Jul 2008, at 13:49, Ian Horrocks wrote:
>> I'm not sure what you mean by not having annotation axioms?
> That was garbled, wasn't it. I meant annotations on axioms, which  
> we do have.
>> We do have annotation axioms, but only for entities and anonymous  
>> individuals.
> And for axioms. I misread and thought the proposal was for getting  
> rid of annotations on axioms. *That's* a non-starter for  
> manchester. I need to investigate more on metaannotations.
>> No one, AFAICT, is suggesting removing them. I believe that what  
>> was said, by Boris, is that we shouldn't implement his earlier  
>> suggestion to resolve issue-16 by making all annotations be  
>> separate axioms, because axiom annotations would then require an  
>> axiom (the one to be annotated) within an axiom (the annotation  
>> axiom) -- a potential problem for the RDF serialisation (at least).
> Yes. Well, I've heard several people over the past year ask for  
> meta-annotations (I believe NCI and Deb are among them. I'm unclear  
> whether the cost benefit is worth it.

The cost of Boris's proposal is the need to embed axioms within  
axioms and issues related to determining if the embedded axiom is  
asserted in the ontology or not. This could be particularly  
problematical in the RDF serialisation where, as we know, there is no  
mechanism for grouping triples, and where it might be difficult to  
distinguish such an embedded axiom from a "bare" axiom.

The cost of Peter's proposal is that we would need arbitrarily nested  
reification. This seems undesirable given that "simple" reification  
is already controversial/problematical -- although Peter doesn't seem  
to see any problem with this (apparently taking the view that if we  
solve the basic problem, then nesting shouldn't introduce any  
additional difficulty).

After discussing both proposals on the teleconf the prevailing  
opinion was that the added complexity of the RDF mapping wouldn't be  
justified -- Boris withdrew his proposal -- but no firm decision was  
taken: we should give Peter a chance to say why he thinks his  
solution is OK and others to make a case as to why is annotations on  
annotations is an important feature.


> Cheers,
> Bijan.

Received on Thursday, 10 July 2008 20:04:46 UTC