Re: Connecting multiple fragment selectors with individual bodies

Hi,

sorry, but could you please give a hint where I could find some more 
background information regarding all these concepts you are discussing ?

What I understand, is that CompositeAnnotation and AnnotationSet are 
concepts to relate multiple annotations, and that these concepts are 
much more suitable than hasSemanticTag to link together the annotation 
elements defined for my use case.

But, what is bothering me is that I have to create a lot of unnecessary 
annotation instances in the model. Moreover, most of these instances 
will not represent annotations themselves, but only dependencies (or may 
be better said attributes) of annotations. Also, these attributes will 
never be intended to become independent "annotations".

What about my proposal to allow targets or at least target specifiers 
having a hasBody property ?
That way, annotations may compile a list of target specifiers with 
fragment selectors, and assign a single, individual body resource to 
each of them. Anything would be fine, at least with regard to my use case.
We could avoid the mass instantiation of unnessary annotations (we 
currently have annotation types which may include up to 15 annotation 
elements, what an overhead regarding the creation of annotation 
instances) and I guess, this would also contribute to a much better 
query performance when several thousands of annotations will be stored 
in the model.

best regards
Lutz


> Rob,
>
> On Mon, Aug 13, 2012 at 4:13 PM, Robert Sanderson <azaroth42@gmail.com 
> <mailto:azaroth42@gmail.com>> wrote:
>
>     Hi Paolo,
>
>     On Mon, Aug 13, 2012 at 9:38 AM, Paolo Ciccarese
>     <paolo.ciccarese@gmail.com <mailto:paolo.ciccarese@gmail.com>> wrote:
>     > Rob,
>     > to me, what Lutz is doing looks more linking separate annotation
>     with a
>     > semantics that, however, in his example is not very explicit. It
>     looks, to
>     > me, very similar to what Kevin was doing with the annotations
>     based on other
>     > annotations.
>
>     Agreed, though I was going from Lutz's response to my initial question
>     where he says:
>
>     In that case, it makes no sense to have 5 non-related annotations for
>     that kind of annotations, because the annotation is only valuable, if
>     all of the related elements can be expressed within a single
>     annotation !
>
>     So I think that both approaches could be valid.
>
>     > I see grouping the annotation with the Composite Annotation as
>     orthogonal
>     > and therefore as possible on top of what Lutz is already doing.
>
>     Agree that linking annotations and grouping them are somewhat
>     orthogonal, hence my red 'x:rel' placeholder relationship between the
>     annotations in the diagram.
>
>
>     > Just to give a little more background about the grouping of
>     annotation, Rob
>     > and I had a side conversation and we see three different levels of
>     > aggregations:
>     > - Annotation Set (by topic, purpose....)
>     >    - Composite Annotation
>     >       - Annotations
>     > I am assuming I could also nest additional levels. Like a
>     > CompositeAnnotation hasAnnotation another CompositeAnnotation.
>
>     Where a set of annotations is collection created for some reason and
>     the individual annotations can be re-aggregated into other arbitrary
>     sets, but a Composite Annotation is a construction for maintaining all
>     of the annotations together, and hence the annotations that compose it
>     should not be disaggregated.
>
>
>
> So if I have:
> Annotation1
> basedOn Annotation2
> basedOn Annotation3
> and
> Annotation2
> basedOn Annotation4
> basedOn Annotation5
>
> If all the annotation tasks have been performed in the same *context* 
> I would probably create one single CompositeAnnotation that lnks all 
> all of them. If I created the second part first - forgive the 
> numeration - I could have a CompositeAnnotation1 linking Annotation2, 
> Annotation4 and Annotation5. Then a CompositeAnnotation2 linking 
> Annotation1, Annotation2 and Annotation3. As we know that Annotation2 
> is based on Annotation4 and Annotation5 because of the relationships, 
> we can now rebuild the hierarchy.
>
> Is that what you mean?
>
> Paolo