Closed bsara closed 3 years ago
The OWL axiom file (e.g. sct2_sRefset_OWLExpressionSnapshot_INT_20210131.txt
) contains the OWL expressions. These hold the stated form of the concepts. These expressions are used to create the OWL ontology file that this toolkit can produce.
OWL uses class definition statements to represent a child -> parent relationships rather than an Is a
attribute. For example the follow axiom owl expression represents the model of the 404684003 |Clinical finding (finding)|
concept and the fact that 138875005 |SNOMED CT Concept (SNOMED RT+CTV3)|
is the only stated parent:
SubClassOf(
:404684003 |Clinical finding (finding)|
:138875005 |SNOMED CT Concept (SNOMED RT+CTV3)|
)
The relationships file (e.g. sct2_Relationship_Snapshot_INT_20210131.txt
) contains the "inferred form" of the concepts. Because of the format of the relationships table it can only hold triples (source, target, type). Here we use the Is a
type relationship to express that a concept is child of another. The relationships in the inferred form come from the classification process which includes OWL reasoning and takes the stated form as input.
thank you so much for the explanation, it is super helpful. (and for the super fast response too)
@kaicode It does make me wonder, however, why are these attribute classes like Is a
included in the resulting owl dataset if they are not used/needed?
I agree that is-a is not needed because it can not be used in owl. It's hard to be so sure about the rest.
I'm not seeing the relationships in the resulting OWL file that I see in the relationship files in the Snomed zip. For instance, I don't see the use of
116680003
(which is theIs a
attribute) anywhere in the resulting OWL file. Am I supposed to not see any of that stuff in there? I'm very new to this toolkit and am not entirely sure that I'm even using the tool correctly.