[topicmapmail] Reification and Type Instance

[Bernard Vatant]

Hi folks

FWIW (2 cents)

In Mondeca ITM data model (which is as you know topic-map-ish withouth 
being conformant to TMDM) we have kept all meta-classes strictly 
disjoint : "topic type", "association type", "role type", "occurrence 
type". So in particular an association type and the matching reification 
topic type must be distinct. But they are linked internally, by a 
"reification class" occurrence put on the association type (arbitrary 
choice -could be by any other convenient mean, such as an ad-hoc 
assocation type). The user interface(s) can display reified associations 
"as if" they were topics. If you export to RDF, they are merged in a 
single class. But under the hood things are kept distinct.
I keep thinking this is clean both technically and conceptually, 
following the good principle "If you are not sure it's ambiguous, then 
it is." You've always better keeping things distinct under the hood and 
present them to interfaces as if they were the same, than the other way 
round.

Bernard

Steve Pepper a écrit :
> | * Robert Cerny
> | >
> | > I have a question regarding the TMDM: If i reify a typed statement,
> | > is the resulting topic an instance of the type of the statement?
>
> * Lars Marius Garshol
> | 
> | This came up a couple of times in ISO meetings, but was never really
> | resolved, partly because in the contexts where it came up, it didn't
> | need to be.
> | 
> | ...
> | 
> | I can't really find any good reason why it shouldn't be the same
> | type. If anyone can I'd be interested to hear it. (Borislav?)
>
> I certainly can't, but...
>
> I have long felt this to be an unresolved issue that ought to be addressed,
> and I was hoping that the "correct" answer would fall out of the work on the
> mapping from the TMDM to the TMRM.
>
> I agree with Lars Marius that both the association and the topic that reify
> it represent the same thing and should therefore be instances of the same
> class. The only thing that worries me is the possible implications for
> existing tools and applications.
>
> Assuming we decide to state explicitly that the type of the association
> becomes the topic of the reifying topic, a number of questions arise:
>
> * Does the inverse apply? That is, does the topic type of the reifying
>   topic become the association type of the association? If so, what
>   happens if the topic is an instance of multiple types (since an
>   association type cannot have multiple types)? In particular, what
>   happens when there are type hierarchies? (Note: I would have a lot
>   of use for the equivalent of rdfs:subpropertyOf.)
>
> * Is the topic map author required to express the type-instance
>   association, between the "association type topic" (e.g. employed-by)
>   and the reifying topic, explicitly? (This would break backward
>   compatibility with existing documents.)
>
> * Is a Topic Maps processor required to instantiate the (implicit)
>   type-instance association explicitly? (This would break backward
>   compatibility with existing processors.)
>
> * Do TMQL and TMCL processors have to infer the association? (This
>   needs to be resolved PDQ, before these standards go to FCD.)
>
> Steve
>
>
> _______________________________________________
> topicmapmail mailing list
> topicmapmail at infoloom.com
> http://www.infoloom.com/mailman/listinfo/topicmapmail
>
>
>   

-- 

*Bernard Vatant
*Knowledge Engineering
----------------------------------------------------
*Mondeca**
*3, cité Nollez 75018 Paris France
Web:    www.mondeca.com <http://www.mondeca.com>
----------------------------------------------------
Tel:       +33 (0) 871 488 459
Mail:     bernard.vatant at mondeca.com <mailto:bernard.vatant at mondeca.com>
Blog:    Leçons de Choses <http://mondeca.wordpress.com/>