Class disjointness in Data example

[gezever]

https://github.com/w3c/sdw-sosa-ssn/blob/gh-pages/ssn/rdf/examples/ACME-Beer/Beer-Shipping-IBS-TH2.ttl contains two individual members of disjoint classes.

An observation cannot be an entity

1.

<67a/observation> a sosa:Observation, prov:Entity ;
    rel:own <acmeBreweryCo> ;
    rel:own <acmeDelivery> ;
    sosa:observedProperty <10/PO202402/21/0001/BeerTemperature> ;
    sosa:hasUltimateFeatureOfInterest <10/PO202402/21/0001/acmePorterSixPackBeerSample> ;
    sosa:madeBySensor <12345/TemperatureSensor>  ;
    sosa:hasFeatureOfInterest <10/PO202402/21/0001/acmePorterSixPackAirSample> ;
    sosa:resultTime "2024-02-22T04:15:05Z"^^xsd:dateTime ;
    sosa:hasResult  [
      a qudt:QuantityValue ;
      qudt:hasUnit unit:DEG_C ;
      qudt:value 19.0 ;
    ] ;

2.

<67b/observation> rdf:type sosa:Observation,prov:Entity ;
    rel:own <acmeBreweryCo> ;
    rel:own <acmeDelivery> ;
    sosa:observedProperty ex:airRelativeHumidity ;
    sosa:hasUltimateFeatureOfInterest <0001/ProductPackaging>;
    sosa:madeBySensor <12345/HumiditySensor>  ;
    sosa:hasFeatureOfInterest <10/PO202402/21/0001/acmePorterSixPackAirSample> ;
    sosa:resultTime "2024-02-22T04:15:05Z"^^xsd:dateTime ;
    sosa:hasResult [
      a qudt:QuantityValue ;
      qudt:hasUnit unit:PERCENT ;
      qudt:value 74 ;
    ] ;

=>

sosa:Observation   rdfs:subClassOf   sosa:Execution. 
sosa:Execution  rdfs:subClassOf   prov:Activity.
prov:Activity  owl:disjointWith  prov:Entity

[gezever]

<123/TruckObserver> a prov:Activity ;
    rdfs:comment "Truck onboard monitoring system" ;
    prov:used <acmeTruck> ;
    prov:generated <67a/observation> ;
    prov:generated <67b/observation> ;
    prov:startedAtTime "2024-02-22T04:15:05Z"^^xsd:dateTime ;
    prov:endedAtTime "2024-02-22T05:55:38Z"^^xsd:dateTime ;
    prov:wasStartedBy <driver> ;
    prov:atLocation <gpsLocation> .

An activity cannot generate an observation.

[dr-shorthair]

Indeed. A prov:Activity generates a prov:Entity which is a disjoint class with Activity.
Ensembles of activities and executions are managed in Collections but this is not fully worked out in SSN.
See the brief mention here: https://www.w3.org/TR/vocab-ssn-2023/#Collections-extensions

[rob-metalinkage]

we should look at sosa:hasResult as a prov:generated sub property then. I might create some SHACL rules in my implementation to generate the PROV model from the implicit model.

[dr-shorthair]

Its already there https://www.w3.org/TR/vocab-ssn-2023/#PROV_Alignment_property

[dr-shorthair]

Missing from PROV is an Activity/subActivity structure.
I added this in the Project Ontology, which is mentioned in the NOTE in https://www.w3.org/TR/vocab-ssn-2023/#Collections-extensions

[gezever]

On the question of sub-activities in PROV: this was intentionally left out of the PROV model.

As discussed in the literature:

Requirement VI8 – No SubActivity:
It is not a requirement of PROV to specify a notion of subactivity. Instead, the Provenance Working Group suggested that relations such as dcterms:hasPart could be used by applications to model subactivities; applications would be responsible for ensuring consistency with the PROV model.

Source:
https://www.sciencedirect.com/science/article/pii/S1570826815000177

Separately from that discussion, I do see the value of the ExecutionCollection pattern introduced in the newer SOSA/SSN specification for grouping executions/activities.

This is conceptually similar to the role that qb:ObservationGroup plays in the W3C Data Cube vocabulary.

[dr-shorthair]

Thanks for the SubActivity pointer.

I see from the text in the paper that part of the reason for not including SubActivity was lack of clarity on the use-case.
Perhaps my Project Ontology, fills that gap a bit.
A 'project' is an obvious application for prov:Activity and projects clearly have sub-activities, etc etc.
TBQH I found the PROV examples rather limited, focussing on research papers and data - information artefacts - while the process-flow model is so much more generally applicable.
In another forum I looked at Provenance of Real Things

I'm a bit reluctant to use dcterms: directly for object-properties. AFAIK there is no OWL encoding of Dublin Core.

[oldskeptic]

I appreciate you bringing this up @gezever . Note that the ACME example and the 7.1 PROV Alignment Module are both "non-normative" and the alignment "sosa:Execution rdfs:subClassOf prov:Activity" is not imported in the example.

I will fiddle with the example some more to avoid breaking if the alignment is imported.

I agree with @dr-shorthair about dcterms:hasPart, it is likely the most abused term in RDF and not in OWL. My current practice is to use the Time Ontology ontology and use time:intervalDuring to telescope things together.