Re: cwm or euler as ontology translator??

   Date: Thu, 26 Jun 2003 07:50:38 -0700
   From: "Jones, David H" <david.h.jones@boeing.com>

   Has anyone looked at the possibility of using cwm or euler as an
   ontology translator?  Such a capability could be used in any
   environment when processing elements don't share the same
   vocabulary, such as a society of agents.

An alternative is to use an FOL theorem prover, which isn't limited to
manipulating triples.  Check out the OntoMerge translator, developed
here at Yale:

http://phd1.cs.yale.edu:4040/ontoMerge.html

It's an ontology-translation service with a modest library of _merged
ontologies_ (or sets of bidirectional translation rules, if you prefer
to think of it that way).  For instance, the following URL

http://cs-www.cs.yale.edu/homes/ddj/ontologies/drc_bbn_gen_merging.pddl

points to a set of axioms relating two different genealogy ontologies.

   The translator would use a series of mapping statements to
   translate triples from one ontology to another.  The translation
   mechanism should support:
   - Transformation based on arithmetic operators, functions. 

Via procedural attachment, we can do that.

   - Lookup 'tables' containing tuples; Used to convert values of one
     property to values of another.

Not clear what you mean.  Is a special technique required here?

   - Mechanism for filtering instances so that translation only
     applies to a subset of instances.

We would require a separate filtering step to extract the dataset you
really care about from a larger dataset.  However, our philosophy is
that it is quite tricky to discard just the right amount of an
ontology or dataset, so we'd rather translate the whole thing.  You
can then do filtering on the output.

-- 
                                             -- Drew McDermott

Received on Friday, 27 June 2003 16:06:14 UTC