XSD issues

[MichalMed]

As the model and specification is ready, XSD file is used to create sample XML files. In this issue is kept the list of issues, solely to determine all of the issues. After that, we'll set up separate issues for them and add priority.

[MichalMed]

Codelist elements shall have iris as an attribute. Codelist labels is string (shall be langString)

 <description>
        <description_text>This is an nice description of dataset.</description_text>
        <description_type>
            <iri>https://vocabs.ccmm.cz/registry/codelist/DescriptionType/abstract</iri>
            <label>Abstract</label>
        </description_type>
    </description>

could be

 <description>
        <description_text>This is an nice description of dataset.</description_text>
        <description_type codelistValue="https://vocabs.ccmm.cz/registry/codelist/DescriptionType/abstract">
          <label xml:lang="en">Abstract</label>
        </description_type>
    </description>

or even

 <description>
        <description_text>This is an nice description of dataset.</description_text>
        <description_type codelistValue="https://vocabs.ccmm.cz/registry/codelist/DescriptionType/abstract">Abstract</description_type>
    </description>

[MichalMed]

Dataset IRI shall be attribute of dataset (IDK if this is possible in Dataspecer).

[MichalMed]

xml:lang atribut při použití langString je v XSD povinný. Měl by být asi spíše volitelný (primární jazyk metadat je popsán v položce metadata record -> language)

[MichalMed]

V případech některých labelů nedává jazyk smysl, například aplikační profil má label většinou technického charakteru (ISO 19115-1:2014/Amd 2:2020)

[MichalMed]

Time instant shall be type used both in TR and time interval. It is like this in conceptual model, but in structural schema they are two separate entities (althought with the same properties).

This is something that probably cannot be done in DataSpecer - to reuse an existing data type in a choice.

Moreover, in the current implementation, even the time instant representing beginning or end of interval must have date type, which does not make sense, because the time interval itself has date type. (solved with ee687e7)

[MichalMed]

order of elements in some cases does not make much sense, e.g. subject definition before title

[MichalMed]

Coar access rights does not have czech version -- can be arranged

[MichalMed]

Related resource vocabulary is not sufficient for description of legislative documents or metadata.
Relation type is not sufficient for description of relations to related resources such as physical sample, tool or legislation document.

[mesemus]

All elements inside the schema should use the same namespace, not a namespace per element. For example:

xmlns="https://schema.ccmm.cz/1.0/datasets" (semantic versioning)

• or -

xmlns="https://schema.ccmm.cz/2025/datasets" (w3c style versioning)

[akastoss]

URLs for identifier schemes should be taken from wikidata, ex. https://www.wikidata.org/wiki/Q51044 for ORCID

[MichalMed]

All elements inside the schema should use the same namespace, not a namespace per element. For example:

xmlns="https://schema.ccmm.cz/1.0/datasets" (semantic versioning)

* or -

xmlns="https://schema.ccmm.cz/2025/datasets" (w3c style versioning)

OK, I'll go with targetNamespace="https://schema.ccmm.cz/research-data/1.0.0" for all schemas

[MichalMed]

URLs for identifier schemes should be taken from wikidata, ex. https://www.wikidata.org/wiki/Q51044 for ORCID

Not sure what you mean. Give here example from metadata.