Re: OWL-cool

On April 22, Enrico Franconi writes:
> 
> On 22/4/03 3:55 pm, Ian Horrocks wrote:
> > The good news is that it is possible to add keys to these langauges
> > without loosing decidability for key inference problems. The bad news
> > is that computational complexity is rather high: it is easy to see
> > that keys are at least as expressive as nominals (oneOf), as integer
> > keys can be used to simulate an arbitrary number of nominals.
> 
> This is not true in general. As the paper cited by Ian says, it is only true
> under the (expensive) assumption that the keys operate over a concrete
> domain. If you want to reason with keys that operate over the abstract
> domain (which to me makes more sense in an abstract conceptual modelling
> context), then the algorithmic complexity is definitely lighter, and easily
> encodable in the available implemented DL systems (without nominals). The
> problem reduces to the efficiency of implemented systems with cardinality
> and inverses.  

I had assumed that we were talking about datatype (concrete domain)
keys - the (archetypal) example in Bob's original email used SSN. The
problem with OWL DL/Lite is that the set of datatype properties is
disjoint from the set of inverse-functional properties (a deliberate
design choice in order to avoid the complexity of keys with
datatypes), so the cardinality+inverse solution doesn't work in this
case - even if one could live with the apparent awkwardness of the
syntax.

Ian

>  
> ciao
> -- e.
> 
> Enrico Franconi                  - franconi@inf.unibz.it
> Free University of Bozen-Bolzano - http://www.inf.unibz.it/~franconi/
> Faculty of Computer Science      - Phone: (+39) 0471-315-642
> I-39100 Bozen-Bolzano BZ, Italy  - Fax:   (+39) 0471-315-649
>

Received on Wednesday, 23 April 2003 11:25:21 UTC