opengeospatial / om-swg

10 stars 6 forks source link

Object of Interest #21

Closed KathiSchleidt closed 3 months ago

KathiSchleidt commented 5 years ago

Created 2 sketches to better explain the CRP: http://ogc.standardstracker.org/attachment.cgi?id=48

ObservedProperty does not allow to separate the observable property from its object of interest For example : for "massic concentration of NO3-", observableProperty = massic concentration & object of interest = NO3-

As a result in 10-004r3, in "observedProperty" registries we can end-up with a long list of every objectOfInterest concatenated with all its 'obervable properties'. For NO3- "object of interest" other observableProperties would be 'molar concentration', 'ppm'

To avoid exchanging this using om:parameter a dedicated property would be useful. This would also allow APIs to filter on it.

Example definition for an "objectOfInterest" property: "This vocabulary contains a set of substances and taxa which may appear as the subject (identified object) of some observable properties." (from"CSIRO Register: substances and taxa" 'http://registry2.it.csiro.au/def/object') also instantiated here ("BRGM Registre: Objets d'intérêts - Paramètres et substances" https://data.geoscience.fr/ncl/_ObjParSubs) with a definition in French (~ the English one)

With regard to the cardinality om:objectOfInterest (0,*). Supporting Use case

Source: http://ogc.standardstracker.org/show_request.cgi?id=608

dr-shorthair commented 5 years ago

Indeed, you can get a combinatorial explosion of observed properties when you consider all the chemical-species, organism, sensor etc. That's why NVS P01 has >30k entries! Having a predictable design pattern for compound-observables is useful here.

Another solution is fuller use of the O&M model itself - e.g. the feature of interest is 'Nitrogen in the water column' ... (though arguably that just moves the same problem to a different place as each feature must get combined with every chemical substance, etc ...) So I would be cautious about adding more properties like 'object-of-interest', since this information already inheres in either the feature-of-interest or observed-property.

Ultimately the best solution is likely to depend on the application and customary practice in a community.

sgrellet commented 4 years ago

@ilkkarinne : ok to also be assigned to that issue

KathiSchleidt commented 4 years ago

I'm starting to see a parallel with the UltimateFeatureOfInterest #3 Is this actually the UltimateObservedProperty?

sgrellet commented 4 years ago

After discussions in Toulouse and webconfs, there is a concensus from a a domain point of view to have this information. The question that remains is how to model this. Various approaches are on the table. We have to discuss the proper mechanics to do so : cross-community and allowing communities to refine this based on their customary practices

sgrellet commented 2 years ago

Issue mainly taken on board by RDA I-ADOPT work. See : https://www.rd-alliance.org/groups/interoperable-descriptions-observable-property-terminology-wg-i-adopt-wg