Description
Public services are defined and provisioned at different levels of granularity and versions. The same generic public service, defined for example at national level, can be delivered in different variants by service providers at the regional or locla levels so on. The generic public services and their versions can differ in terms of descriptions, delivery channels, costs, outputs, evidences, criteria, etc.
Members of the CPSV-AP working group has raised the need of modelling this relationship between generic public services and their versions.
Proposed solution
We propose to include a new reflexive property on Public Service of type subclassOf. It would mean that a Public Service can be specialised into a public service version. All properties and relationships will be inherrited, and may be overwritten if they differ.
Comments
I'm in favour of the second option. I'm not sure whether the spatial property will be enough to distinguish the level of abstraction. What if you've got two public services where one is a specialisation of the other while they both have the same spatial 'applicability' (but eg the terms of service delivery may vary depending on the abstract/concrete nature). As also is mentioned, the relationship 'related' is too generic in this use case.
Not sure which 2 options Thomas is talking about, as I see only one suggestion in the original post.
I have some concern about the "a new CPSV-AP property of type subclassOf" statement. I'd like to point out that the proposed attribute should NOT be a specialisation of subclassOf, since that is meant to link classes, rather than instances, to each other.
I agree with Dieter. Moreover, we should be very careful in distinguishing subclass from versioning. The subclass mechanism is already built in in UML, so to speak. At a national level we can have a class that describes a particular *kind* of public service, which would be a specialization of cpsv-ap:public service. In turn, this class can be specialized by different "variants" of such service kind.
Thinking at the token level of *the same* service that has different variants is a different story. I am not sure if we need this...
I remember reading two options here (one on versioning through the spatial property, thus starting from the same PS but denoting variants depending on their spatial applicability and one that's still here).
Firstly, it has to be clear we're talking about types of public services that essentially aim at the same goal and are interrelated. The clue is that the different but very alike public services each have their own right of existence, but we need to be able to relate them.
Eg. parliament decides each municipilaty may fund their inhabitants in placing a cistern (where the municipality gets funding if the apply). This implies a generic description stating the decretal facts about the public service. But also n local descriptions, each with their finesses. How do we link those?
A subclass may indeed not be the best solution - I was thinking in terms of a specialization of the relationship 'related' as a circular relationship within publicService-class. (thus a third relationship)
Thomas' example makes things a bit clearer for me. One thing isn't clear to me yet, however: is the federal-level service also usable by citizens or not? If it is not, I would argue this is not a valid service (and therefor falls out of scope of the current available elements).
Regardless of the above example, this reminds me a bit of the FRBR ontology, though I haven't thought of the details yet. (Wiki link)
During the webinar held on the 12th of May, it was agreed with the WG to include this relationship as an association between public services. The changes will be included in the new draft version of the CPSV-AP.
The addition of the isSpecialisationOf property will be put on hold as agreed by the WG during the webinar held on the 9th of June. In the meantime, we will analyse the different implementations that the MSs used to cover the relation between generic and specialised public services.