This is very interesting!! Alan I totally agree with you that your goal is
definitely a good one to have.
[Alan] > Put another way, the goal might be stated as wanting to get both
*all* available answers to our questions, and *only* correct answers
to our questions, and both the above contribute to achieving that goal.
Usually soundness is used to convey correctness as is used in data modeling.
Soundness is achieved in a schema after it is designed and run through the
integrity design process, that is, a mathematical set of steps that ensure
"soundness". Once a schema is correctly designed one can ensure that no two
conflicting answers can come out from the same query.
In the SW domain, the power of RDF is the easiness of adding RDF statements
with out a design process. Two individuals can add two statements that
happen to be conflicting. Nothing in an ontology would reject the
statements as long as the syntax is correct.
Hence, to get *all* available answers does not guarantee soundness, as two
RDF statements could exist in one ontology that are conflicting. Now to add
the "and" condition to *only* correct answers is interesting to explore!!
Correct according to whom? Even if some group agrees that these are the
correct rules, how to enforce it?? How to achieve such a goal in a
language versus in a model? ....
[VK] I guess soundness is still the same, don't want "wrong" answers to
be
returned in any case.
Yes, I agree to that. Unless we need to redefine soundness in a new
meaning?
[VK] But completeness would be based on the what's there in the
virtual integrated DB/KB.
Completenes is typically used on a set of operators not data.
Simplistically, If a set of operators can execute *all* the possible actions
to alter a domain -- completness is acheived. I am being very casual here
because mathematicily this is a subject and a theory by itself. I am sure
more people on this group can elaborate on this more precisely.
Cheers
-Wafik
