OWL ontology for unit

[David Berry]

On Mon, 1 Sep 2003, Sebastien Derriere wrote:

> >     This all should prove really useful if we get to the stage where we
> > want to automatically merge columns of
> > data or quantity objects and we need an automated way to check if the
> > two streams of data have the same
> > physical units and then, if they are, determine what is the constant
> > factor between them.
>
>   I think that ontologies can prove very useful to describe some
> semantic links between concepts. However, I think that they are
> not the proper way to describe mathematical relationships. I have
> not been able to find any single application of an ontology to
> interpreting mathematical relations, except the theoretical paper
> from Gruber and Olsen, 1994.
>   I think that ontologies are not meant for this.

Once you have analysed the unit string into its basic SI units and the
relationships between them, it is surely possible to use this analysis to
work out the transformation between them - without the need for any
further ontologes. The Starlink AST library for manipulating WCS
information (http://axp0.ast.man.ac.uk/~dsb/ast/ast.html) does this.

One point, the relationship may not be a simple scaling factor - for
instance, you may have a column in log(Hz) and another in GHz. The
ontology presented by Ed does seem to miss a few functions such as log and
exp - the description of unit handling in the FITS WCS paper I is a useful
summary of the facilities which will probably be needed.

David