> > You can also use RDF reification and old fashioned certainty factors to
> > cross the bridge :
> >
> > [Nancy hasVirus #SARS] hasCertainty 0.01
> >
>
> You could certainly do this, but an OWL engine wouldn't be able
> to make any
> sense out of it.
No, but it provides a means of associating statements with 'analog' style
data.
RDF graphs could represent Bayesian networks but that
> doesn't mean that OWL inferencing engines would somehow understand them --
> its at this precise point that you would need to invoke some as yet to be
> defined -- let alone discussed-- layer of the SW that would
> understand this
> type of logic.
Yep, but I reckon it's easier than you make it sound. Even if the OWL system
is only used for transport/routing of analog data, it's got a lot of
potential. Add a bit of rudimentary fuzzy logic and it should get fun.
Statistical systems can provide (real-)numerical data which can be
represented in OWL which can be pushed around & reasoned with in OWL terms,
pumped out as numerical data to external statistical analysis engines, the
results re-coded as RDF statements...
My bet would be on some melding of any of a number of
> (currently available) statistical processors as input to (currently
> available) logical classifiers, and perhaps this would be an iterative
> process.
Quite.
Cheers,
Danny.
