Lists Home |
Date Index |
m batsis wrote,
> Miles Sabin wrote:
> > Not quite. I'm saying that vocabulary to vocabulary mappings can be
> > expressed in XSLT as well as in OWL, and I'm suggesting that for
> > many practical purposes XSLT is likely to be a better bet.
> Sorry, I have to dissagree. I cannot imagine even a single case where
> that is true. A transformation can only be applied to a "screenshot"
> of the source data. Also, such an attempt may result in loss of
> semantic info. A transformation may result in a correct mapping
> between ontologies, but will fail in producing implicit statements
> that would result from those mappings after applied with
Umm ... this is going round in circles.
The original claim was,
SW technologies are cool. Look! OWL equivalance statements enable
My response was,
Just define the mapping using XSLT (or whatever).
Your response now is,
That's no good, you don't get the equivalence statements.
But I _don't_want_ the equivalence statements: I just want the mapping,
and I really don't care whether or not OWL equivalence statements can
be derived from it or not.
In any case, many completely innocuous looking mappings aren't
expressible in DLs. Suppose I want to map from a vocabulary that works
with height, width and depth to one which works with volume. I'll want
to be able to say that volume = height*width*depth. But unless I'm very
much mistaken that kind of incredibly simple arithmetical constraint is
beyond the expressive capabilities of DLs.
OWL works for you, then fine. But that only supports the claim that SW
technologies are likely to be widely applicable and more effective than
the alternatives if most applications are like yours.
Care to try and persuade me?