Re: References to "application profiles"

On Sun, Aug 28, 2011 at 11:34:58PM +0200, Antoine Isaac wrote:
> Fair enough.
> Note that the confusion between "alignment" and "ontology mapping" is quite understandable: as soon as you include in it similarity between individuals (owl:sameAs or softer), then "ontology mapping" may cover the entire realm of semantic alignment, in the RDF(S)/OWL world.
> Note that as an ontology matching researcher, I am using the terminology from that area:
> - matching = the process of establishing connections (manually or using an automatized technique)
> - mapping = an individual correspondence (e.g., between class A and class B)
> - alignment = a set of mappings between two datasets/ontologies
> 
> So I'd propose to replace
> "which provides methods for mapping equivalences across vocabularies ([http://www.w3.org/TR/2004/REC-owl-guide-20040210/#OntologyMapping ontology mapping])"
> by
> "which provides elements to represent alignments across vocabularies ([http://www.w3.org/TR/2004/REC-owl-guide-20040210/#OntologyMapping ontology mappings])"

I like "represent alignments" better than "mapping equivalences"; actually, the
previous version said "equivalencies" -- a word which AFAICT does not exist.

By "elements", do you mean "properties" (e.g., owl:sameAs)?  Do we feel okay
about calling owl:sameAs an "element" -- something one wouldn't say at a
Semantic Web conference but is arguably more consistent with our use of
"element set"?  If we decide against using "property" because we do not define
it in the report as a synonym for "elements" (though perhaps it should be),
then we would need to fix a reference to "standard library properties and
vocabularies".

> But that's a mere suggestion. Having been confused by all that years ago (and
> still being quite a bit) I understand that you may be unconvinced. Even
> though one big plus of my wording is that it's more compatible with the
> section in the OWL specs, which includes "different from" links -- thus quite
> far from "equivalences" (an alignment can indeed also include mappings that
> denote dissimilarity...).

Good point!

Tom

Received on Monday, 29 August 2011 01:18:24 UTC