search

opengeospatial / om-swg

10 stars 6 forks source link

Multiple Categorization Types? #48

Closed KathiSchleidt closed 4 years ago

KathiSchleidt commented 4 years ago

Pertaining to soft typing of observations we're running into the requirement for multiple categorization types on an Observation. Background is the question of why are observations only specialized by result type, not by the other observation associations, a recent customer request pertained to specializing observations by FoI type, a further use case could pertain to observable properties (singular or grouped) being observed. Is there some good practice for this type of multiple typing? related to #47 Soft Typing

dr-shorthair commented 4 years ago

Multiple types fine by me. Class intersection.

sgrellet commented 4 years ago

So a hint of websem rationale in our UML model. fine with me. would then make sense to keep our AbstractObservation:type attribute named type to refer to rdf:type property

KathiSchleidt commented 4 years ago

Is the Observation Type really the same as being referred to by the rdf:type property? Just checking as not that knowledgeable of the RDF details, if it does would be perfect :)

dr-shorthair commented 4 years ago

RDFS semantics implies that individuals whose rdf:type are related to other resources through rdfs:subClassOf axioms are also members of those super-classes. And resources which can be the object of rdf:type properties can also be the object of rdfs:domain and rdfs:range axioms. So in those senses it is a special property. But in general rdf:type is just another property.

KathiSchleidt commented 4 years ago

Agreed that multiple (Non-conflicting) categories are useful both on Observations and collections