Resolution to OWLish question?

Thanks to everyone who offered suggestions- it forced me to clarify my thinking.  I believe the model that will do everything I want (and because I'm looking for values that are specified per instance, not per class or subclass, I don't think the restriction approach will do) is:

<ClassA rdf:ID="thisClass">
  <hasMapping rdf:resource="#thatClass/>
  <hasMappingClass>
      <Map>
         <hasTarget rdf:resource="#thatClass/>
         <usesJavaClass>org.java.blah</usesJavaClass>
      </Map>
 </hasMappingClass>
</ClassA>

I want the first property so that I can capture the fact that it is transitive (and a subproperty of it is symmetric) at the ontology level.  I can't define ontological transitivity/symmetry in the second property, I don't think, since it does not directly relate two instances of ClassA. And I think the presence of the first property will also enable me to write faster, smaller inference rulesets.

My issue with this model is the redundancy- the mappedto class is essentially specified twice- but I may just have to live with that.

Hope this makes sense...  and thanks to all of you who replied.

Linda