|
|
| Re: [Ontology v0.6] Common & Metadata class [message #3790 is a reply to message #3526] |
Tue, 25 November 2025 09:53   |
Larissa Zhuchyi
Messages: 80
Registered: November 2022
|
Member |
|
|
Dear Mathias
Please excuse my late reply. Issue of having containers in the ontology was discussed 2023-10-16 on the ontology working group meeting. The decision was documented as follows "remove classes inspired by container elements from ontology if they have no additional info". Therefore the issue you are rising in the forum seems to be already supported by the working group. My presentation for the meeting from 2023-10-16 however only mentioned elements of XSD having no attributes i.e. "no additional info".
Since the establishing of the ontology working group the development of railML ontology was based on railML 3.2 gradually extending to railML 3.3. As soon as <common> in railML 3.2 and 3.3 has attribute @id, this element was not recognized as container during railML ontology development.
This issue and the one of redundant attribute infrastructure/@id was acknowledged by railML coordinators during development of railML 3.4. As you can see at https://development.railml.org/railml/version3/-/tree/3.4-be ta2?ref_type=tags <common> and <infrastructure> do not have attribute id anymore. When extending railML ontology to cover railML 3.4 as well, I suppose, infrastructure/@id and common/@id can be marked as deprecated as well as classes "infrastructure" and "common". Does this sound reasonable to you?
Removal of class metadata is not clear to me however. Metadata is supposed to describe source, creator, description of data etc, essentially data management. Since ontology is a set of triples, these properties still need to be attached to an individual or a blank node. How would you manage removal of metadata container in this case? Would you add an individual for each dataset in a similar manner as they do using Provenance ontology?
Sincerely,
Larissa Zhuchyi – Ontology Researcher
railML.org (Registry of Associations: VR 5750)
Altplauen 19h; 01187 Dresden; Germany www.railML.org
|
|
|
|
| Re: [Ontology v0.6] Common & Metadata class [message #3792 is a reply to message #3790] |
Tue, 25 November 2025 19:58 |
Mathias Vanden Auweele
Messages: 85
Registered: February 2025
Location: Brussels
|
Member |
|
|
Quote:When extending railML ontology to cover railML 3.4 as well, I suppose, infrastructure/@id and common/@id can be marked as deprecated as well as classes "infrastructure" and "common". Does this sound reasonable to you?
Yes
Quote:
Removal of class metadata is not clear to me however. Metadata is supposed to describe source, creator, description of data etc, essentially data management. Since ontology is a set of triples, these properties still need to be attached to an individual or a blank node. How would you manage removal of metadata container in this case? Would you add an individual for each dataset in a similar manner as they do using Provenance ontology?
That's a good question. To what do we refer when we add metadata predicates? If we would have the class "Metadata" inside the ontology, then we could create 1, 2 ... as many instances of that class as we want inside one knowledge graph and add same or different objects to it. The problem is that a knowledge graph is not a railML file. It can be many railML files. So for me, having the "Metadata" class inside the ontology would mean to also add a class "railML file" to the ontology. Because the metadata should be related to this file, and not to the knowledge graph.
That feels strange to me. We could think one level up and look at the knowledge graph as a dataset in itself, as a file, and then link metadata (similar to DCAT approach) to the file.
Mathias Vanden Auweele
Railway data freelancer
https://matdata.eu
Brussels, Belgium
|
|
|
|
Members
Search
Help
Login
Home
