Closed ajnelson-nist closed 3 years ago
@ajnelson-nist Yes, this, and many others like it, is why separating these domain assertions the way we resently did was important to do. In the example you cite here any time the observable:emailAddress in used the subject of the triple will be asserted to be of rdf:type EmailAddress AND WhoisRegistryInfoType. It is not logical. Hence, logically inconsistent.
So, since I believe these domain assertions are wrong, adopters like DarkLight should not import the observable-da.ttl file. Or any of the xxxx-da.ttl files, in my opinion. If they do, they cannot use an inferencing engine (reasoner).
This issue will be addressed through implementation of JIRA OC-68: https://unifiedcyberontology.atlassian.net/browse/OC-68 and CP-13: https://unifiedcyberontology.atlassian.net/wiki/spaces/OC/pages/984678401/UCO+CP-23+Convert+current+property+restrictions+and+domain+assertions+to+SHACL+shapes
I stumbled on this while aligning the CASE example
accounts.json
with UCO 0.3.0. This is also true for today's state ofdevelop
.observable:emailAddress
is a property of theobservable:EmailAccount
facet. However, the definition of the property states its domain is the facetobservable:WhoisContactType
; this domain definition was carried intoobservable-da.ttl
. Poking aroundobservable.ttl
, I seeemailAddress
is also a property ofWhoisRegistrarInfoType
.The domain assertion is incorrectly restrictive, and should either be expanded or removed.
This somewhat relates to Issue 25.