Discussion to "Uncertainty Ontology" here
begin P. Vojtas
for discussion my picture of ontology Task_oriented_uncertainty_ontology.pdf
end P. Vojtas
I'd vote that we use 'proposition' in place of 'sentence'. And we might clarify the precise sense of 'proposition' that we have in mind, I like the third one that Peter Suber notes in his FOL glossayr (http://www.earlham.edu/~peters/courses/logsys/glossary.htm#Proposition), i.e., "In logic generally (for some), the meaning of a sentence that is invariant through all the paraphrases and translations of the sentence."
Also, I'm not sure why we'd say "a document that contains at least one sentence, but possibly more" (is that supposed to be a meaning of 'sentence'?) I think this part can be removed.
To be provocative, let me also suggest that propositions are the only things that have uncertainty values, and they link to probabilities via binary relations or triples. A proposition either has an uncertainty value via a direct link, propositions about the future, propositions about the kinds of facts discussed in quantum physics, etc. or we might see uncertainty as a property of an agent and a proposition. (probability (<MikePool, 'William Taft is the 27th president of the U.S.'>, .75).
beginPeter Top ontology
Sentence is the terminology of W3C is a triple (subject, predicate, object)
I have doubts about such a triple in our ontology (Sentence, hasUncertainty, Uncertainty)
main point is who asigned this uncertainty
I argue that assigned uncertainty depends on the agent, his knowledge, point of view
Usualy such a situation is solved by reification - and this is difficult to handle by ontology processors
tower of Babel I sugest to extend our ontology by instances (extend the TBox by ABox), so far our discussion is about meaning of latin / greek words, look to wikipedia for (a possibe) explanation of aleatory and epistemic, I sugest to use english description using anglo-saxon words rather than latin-greek
Francis Fung's comment: Regarding this: (probability (<MikePool, 'William Taft is the 27th president of the U.S.'>, .75). Just to be clear, is this an assertion that MikePool believes with .75 probability that William Taft is the 27th president of the US? (Presumably he has no almanac at hand.)
Disagreement below by Vipul Kashyap
- I view Bayesian networks, Fuzzy Logic, Dempster Schaeffer, etc. as representation/reasoning tools on uncertain information. So the proposal was: let's have two classes in the ontology:
- Models/Techniques for dealing with Uncertainty... Mitch seemed to suggest that these things categorized different types of uncertainty, though in my mind, given a piece of uncertain information, I can use any of the above techniques (possibly inappropriately) to represent and reason with it.
- I prefer to call it excerpt of an ontology about "imperfect knowledge" ...
- I also recommend to use an OWL Editor to define the ontology ...
- There have been many attempts in the past to make such an ontology ... and none was accepted as reference ... so there may be a good reason for that :-((