Closed TortugaAttack closed 3 years ago
Is that the whole ontology? I cannot spot any axioms what would allow inference of individuals belonging to complements of those classes in there (no classes declared as disjoint, no individuals declared as different).
Ah sorry my fault, I was thinking/arguing from a Closed World assumption. I guess it is not configurable in Openllet to use a Closed World assumption?
If I use the
OpenlletReasoner
and want to retrieve any Instances for any Class Expression containing Owl Complements, it either yields no result even though they are definitely Individuals fitting or some bot not all.I created a small Ontology and code for reproducing that, and it turns out that the getInstances always returned 0 elements for the ClassExpressions containing a complement. However using https://github.com/SmartDataAnalytics/SML-Bench/blob/updates/learningtasks/carcinogenesis/owl/data/carcinogenesis.owl and f.e.
not carcinogenesis:Bond-2
contains 13055 Instances, so it is not always empty, but it should be way more.Small Ontology file for reproducing:
Code which shows the problem: