Overly constrained test for completeness of certain individuals (false positive due to punning potential) #195
Replies: 4 comments
-
|
I believe that the current test takes into account the inverse properties. The codes reported by the test for commit c5c097a are not (yet?) used in hasValue restrictions for the inverse properties: |
Beta Was this translation helpful? Give feedback.
-
|
@merelog I think the real issue is to ignore individuals with respect to the properties 'denotes' and 'isSignifiedBy' in this test, as unless we create exhaustive examples for every case, we will have codes defined that are only used by the GSRS or client data, not in our examples, and so I think this is overkill. There will be additional examples that use some of these codes over time, but I think for any code set, there will be some we don't cover in our examples, in other words. |
Beta Was this translation helpful? Give feedback.
-
|
@ElisaKendall I can remove the test, but adding exceptions to it does not make sense from the technical point of view. This will add unnecessary technical debt to the infrastructure and increase its maintenance cost. |
Beta Was this translation helpful? Give feedback.
-
|
@mereolog Ok - well, I think it is useful in cases other than codes, so let's leave it for now with the understanding that codes are false positives. |
Beta Was this translation helpful? Give feedback.
-
In defining the codes for ingredient roles in the medicinal products ontology, we cannot say that a given code denotes a class of roles due to punning. Instead we have a restriction on the class of roles that says that it is signified by the code (using a hasValue restriction).
I would like to see this test incorporate a check for whether or not there is a hasValue restriction on the inverse property using that code, and if so, it is not an incomplete individual.
Beta Was this translation helpful? Give feedback.
All reactions