ISSUE-12

2008-03-08 19:13:06: I agree with Mike's statement on the ontological aspect. However, I don't see Empirical as being an uncertainty type. Empirical is a property directly related to evidence, not to uncertainty.
If a given hypothesis can be completely proved by observing data, then there is no uncertainty left. If I can't collect all data to completely satisfy that hypothesis, then I can resort to a statistical experiment and collect Empirical data that I'd use to (statistically) support my eventual conclusions. If no experiment can be performed to satisfy that hypothesis then I may attempt to prove it via theoretical considerations. In all cases but the first, uncertainty is due to either (or both) my inability to collect complete information (i.e. Incompleteness) or to the aleatory nature of the process being analyzed (i.e. Randomness).
A random process is one that doesn't have a deterministic outcome, but instead it follows a probability distribution. I understand we have addressed random processes in UncertaintyNature: Aleatory, but we may consider that a process which has UncertaintyNature: Aleatory will have UncertaintyType: Randomness.
In other words, I'd "promote" Randomness and prune Empirical from being a sibling of Ambiguity, Inconsistency, Vagueness, and Incompleteness.

[Paulo Costa]

2008-03-08 21:05:46: Actually the diagram does not suggest this. The "isa" link means subClassOf, not instance of. So the diagram states that Randomness is a subclass of Empirical, which then is subclass of UncertaintyType.

If this clarification resolves this issue, then we are done. If not, then I would like to hear more from Mike.
[Mitch Kokar]