Re: Cardinality restrictions (was: comparing DAML-ONT and OIL)

yes - I think there is universal agreement that cardinality per domain is 
useful and desirable.  I personally believe they are much more useful than 
cardinality in a global sense.

There appeared to be historical interest in cardinality in a global sense and 
leaving it there along with adding the domain specific cardinality will give 
us the opportunity to see how it is used in the fullness of time.  
It will however give us additional opportunites for contradictions - having 
both constructs will allow us to say that globally a property may have a 
cardinality of 2  (thus have no more than 2 fillers) but later one could add 
either a minimum cardinality of 3  or an exact cardinality of 3 (thus a 
minimum cardinality and a max cardinality of 3) on a particular domain, thus 
causing a conflict.

I see two courses:
(1)  add cardinality on a per domain basis in ADDITION to keeping cardinality 
on a global basis.  Thus we have a monotonic addition to the expressive power.
(2)  since the release is still so early, take away cardinality on a global 
basis and add it only on a per domain basis

Many of us have spent careers encoding cardinality on a per domain basis and 
have not felt a need for it (other than stating that a role is functional) on 
a global basis so I can easily support option 2.  That is what I would have 
chosen personally.
Given that the other is already there though, the path of least resistance is 
to choose path 1.

Deborah
dlm@ksl.stanford.edu

Received on Sunday, 15 October 2000 01:54:06 UTC