SADI input and output classes

Discussion here of asserted classes, versus classes defined as property restrictions.  SADI works best when you define your input and output classes based on what properties input data must have, versus what Class it must be.  Thinking about it from the user-perspective - your users have been assembling their database from a variety of third-party sources unknown to you; you cannot predict where they got their data properties, or from which classification systems they were derived.  As such, you cannot predict what ontological classes they might have in their dataset.  But you shouldn’t care!  So long as their database contains individuals who have the data-properties that YOU require to execute your analysis, those individuals should be valid input to your service.  This is why, in SADI, we strongly recommend that you define your input OWL class in terms of property restrictions, rather than asserting which ontological class you require.