search

HyperAgents / hmas

An ontology to describe Hypermedia Multi-Agent Systems, interactions, and organizations.
https://purl.org/hmas/
1 stars 0 forks source link

"directlyContains" dangerous and possibly not needed #62

Closed Antoine-Zimmermann closed 2 years ago

Antoine-Zimmermann commented 2 years ago

Whenever C1 contains C2, it is not necessarily clear whether C1 hmas:directlyContains C2 or not. So it could be tempting to use hmas:transitiveContains every time to avoid trouble. Or, conversely, it can be tempting to use hmas:directlyContains because one does not envision an intermediary subsystem. But not envisioning such relation does not mean there is none. After examining more carefully the situation, it may appear that there is after all an intermediate subsystem that strictly contains C2 and is strictly contained in C1.

Besides, there is an argument that this mimics skos:broader/skos:broaderTransitive but skos:broader does not imply "directly broader". We can simply have hmas:contains and hmas:transitivelyContains, with the subproperty relation between the two.

DrLeturc commented 2 years ago

hmas:transitivelyContains should never be used and only hmas:directlyContains can be used. The actual comments of hmas:transitivelyContains says :

      `rdfs:comment "A relation that refers to all logical containers in which a resource is contained. (Never instantiate this class)"@en ; `

It should be better explained in the definition of those properties and maybe it is not clear why we forbid to use hmas:transitivelyContains . Maybe we shoudl replace the current comments with :

rdfs:comment "A relation that refers to all logical containers in which a resource is contained. This class is used to get all indirect inclusion relations between workspaces and subworkspace. This relation should never be instantiated. "@en ;

Concerning using skos:broader/skos:broarderTransitive, skos:broader is not transitive but it entails skos:broarderTransitive as the transitive property we want to obtain. We wanted that everything which is contained in its workspace A is transitivelyContained in all workspaces W where A is a subWorkspace of W. skos:broarder allows to do that. So we could consider hmas:contains as a subPropertyOf skos:broader in order to specify that our contain relation is a particular relation where the domain is a hmas:Workspace and the range is a rdf:resource. Furthermore, it would remove the temptation to use a hmas:transitivelyContains if we remove this property from the ontology. skos:broarderTransitive will be used to get all the deductions between the inclusion relations of workspaces.

FabienGandon commented 2 years ago

This is linked to issue #39 and the possibility of declaring hmas:directlyContains as Asymmetric if we want to drop that checkpoint we can have hmas:contains/hmas:transitivelyContains just like skos:broader/skos:broaderTransitive.

@Antoine-Zimmermann : I believe we have to be ready for changes to happen: for instance at any time a user may want to change the inclusion order between two containers, introduce new ones, etc. but to me this just mean that at a given time the model must be coherent not that two versions of the model must be coherent.

@DrLeturc : I don't get the sentence "we could consider hmas:contains as a subClassOf skos:broader in order" these are properties and I wouldn't use SKOS here as we are not modelling concepts.

DrLeturc commented 2 years ago

I meant to redefine skos:broader in the context of hMAS with hmas:contains by using rdfs:subPropertyOf.

DrLeturc commented 2 years ago

In the description of skos:broarder: A hierarchical link between two concepts indicates that one is in some way more general ("broader") than the other ("narrower").

Seeing a "super-workspace" as a kind of more general concept doesn't shock me e.g. a "super-workspace" isBroader (in some way) than a "sub-workspace" that is contained in, and which contains a subset of resources.

But can we say the same for the rdf:resources that are contained in a :workspace e.g. using hmas:contains from a hmas:Workspace to hmas:Artifact or hmas:Agent ? Mmmh.. maybe not.