dr-shorthair
commented
3 years ago It is precisely because of the difficulty of correlating multiple results with multiple observed-properties that it was decided to model this as one-result per observation. Sure you could do clever things with compound or composite observables and results, but specifying that is out-of-scope for SSN/SOSA. IMAO it would have to be (could only be?) implemented as a community practice.
It is possible to observe more than one property of a feature at the same time. This may be handled with complex results, but it is unclear how a complex result can be mapped to more than one observed property.
the text at https://www.w3.org/TR/vocab-ssn/#SOSAhasResult for hasResult is sufficient open to allow any possible implementation via the "representing the value associated with" wording
"Relation linking an Observation and a Sensor or Actuator and a Result, which contains a value representing the value associated with the observed Property."
if the one result contains values for more than one observable property of the FOI then this restriction makes it impossible to make the associations:
sosa:observedProperty EXACTLY 1
the definition of ObservableProperty is quite loose however, so it could possible be implemented as a sequence of atomic RDF properties perhaps?
At any rate, some guidance on this seems appropriate, as its quite hard to apply SOSA to existing Features that dont have the exact complex result containers, but observations need to bundle a range of results. Possible a ObservationCollection is the only option, but a special type where the collection is a set of related results sharing a common observation event ?