julianrojas87 / era-vocabulary Goto Github PK

Currently IRI: http://era.europa.eu/ns#inCountry is defined with the intended use

Indicates the country in which the entity resides.

    rdfs:label "in country"@en ;
    rdfs:comment "Indicates the country in which the entity resides."@en ;
    rdfs:isDefinedBy era: ;
    rdfs:domain [ a owl:Class ;
                  owl:unionOf ( era:OperationalPoint era:VehicleType )
                ] ;
    rdfs:range skos:Concept ;
    dct:created "2020-09-01"^^xsd:date ;
    dct:modified "2020-09-01"^^xsd:date ;
    vs:term_status "stable" .

It has domain owl:unionOf ( era:OperationalPoint era:VehicleType . OP have location rather than residence. For VT the issue the intended use is ambiguous: "manufactured in", "spend most of it's lifecycle in", "has manufacturer with coutry of residence" etc.

Possible solutions: changing the comment, creating a sub-property, changing the domain (OP or Manufacturer).

Consider removing
?class a rdfs:Class

in all cases where ?class a owl:Class

because owl:Class rdfs:subClassOf rdfs:Class

unless there is some specific reason to assert such inferences but if that's the case, it should be done for all classes for consistency.

Currently, the following classes are owl:Class, rdfs:Class

• era:InternalNodeLink
• era:MicroLink
• era:MicroNode
• era:NodePort
• era:OperationalPoint
• era:Track
• era:Tunnel
• era:VehicleType

as discussed

1. These are wrong

era-op-types:OperationalPointType
skos:inScheme era-skos:  .
era-skos: a skos:ConceptScheme

because a ConceptScheme is a single thesaurus and consists of Concepts, not of subclasses of Concept .

It's better to declare

era-op-types:OperationalPointType
rdfa:isDefinedBy era-skos:  .
era-skos: a owl:Ontology

2. each Concept must belong to a ConceptScheme. So in addition to this

era-op-types:borderPoint a era-op-types:OperationalPointType

You better add this

era-op-types:borderPoint skos:inScheme era-op-types:OperationalPointType.
era-op-types:OperationalPointType a skos:ConceptScheme.

I don't think it's a bug problem that era-op-types:OperationalPointType wull become both an individual ( skos:ConceptScheme) and a Class. OWL Punning allows that.

For example: era:otherEnergySupplySystem with era:energySupplySystem

Similarly for:
era:otherAxleBearingConditionMonitoring
era:otherRailInclination
era:otherWheelSetGaugeChangeoverFacility
era:otherSnowIceHailConditions
era:otherPantographHead
era:otherContactStripMaterial
era:otherPlatformHeight
era:otherProtectionLegacySystem
era:otherLegacyRadioSystem

in #v0.9.0

When the development, maintenance and publishing responsibility for all reference datasets is done by one person/team, it is acceptable to keep them all in one file for convenience. But once proper governance is established, they should be split.

In other words, during the pilot phase is fine to keep them like that, but this should be changed right after.

Inherited from the relational database, there is no relationship between classes but we should review and add those if they exist.

Since the domain of the ontology is larger than train vehicles, "category" is ambiguous. The URI may remain but consider changing the label to "vehicle category", after consulting with ERA.

Same of subCategory.

Proposal:
era-lnid:LineNationalID
in #v.0.9.0

(It is an identifier for the whole line, so broader than the SoL or Track)

Describe the procedure to report issues about ERA vocabulary in the README.

This becomes obvious from the different prefix per class.

...which is a contradiction because skos:Concept and skos:Collection are disjoint.

I haven't analysed which declaration leads to this.

In the declaration
owl:imports schema:, era-skos:, eu-countries:, wgs:, geosparql: .
schema and georsparl are fine but era-skos and eu-coutries not.

As suggested in #8, we could add a predicate to refer to the index ID of every parameter, as they were defined originally in ERATV and/or RINF.

The manufacturers should be with URIs etc but they are master data and should be treated as such (to find out if there is register etc). They are legal entities and should not be treated as concepts.

Currently the domain of era:inCountry is the union of VT and OP.

However, the actual use is broader.

1 | era:OperationalPoint
2 | geosparql:Feature
3 | wgs:SpatialThing
4 | era:VehicleType

Apart from that to allow for a higher interoperability and re-use, the era ontology should "learn" but not be coupled with the current dataset.
Suggestion: remove domain declaration and for orientation list in the note the typical expected domain classes, based on the current data set.

Currently:
Author = https://julianrojas.org/#me, therefore the Cite as = https://julianrojas.org/#me. ERA vocabulary. Revision: v0.9.0.

Modify to:
Author = European Union Agency for Railways, therefore Cite as = European Union Agency for Railways. ERA Vocabulary. Revision: v0.9.0