Currently, the refinement operator based algorithm in DL-Learner takes the subsumption hierarchy of concepts into account, but not the role/property hierarchy. One could extend the core algorithm to cover this, which would enable more efficient learning in cases where a hierarchy of roles/properties exist.
Logged In: YES
user_id=1620101
Originator: YES
The refinement operator based algorithm has been extended to take care of role hierarchies.