Provenance

[rweigel]

We want to express

1. "Dataset B is a revision of dataset A" in the sense that B is a better version of A. Similar to "manuscript B is a revision of manuscript A" because typos were corrected.

2. "Dataset B1 is based on dataset A" and "Dataset B2 is based on dataset A". B1 and B2 are not equivalent, but are closely related. An example is

• dataset B1 was created from dataset A by removing spikes with automated algorithm 1
• dataset B2 was created from dataset A by removing spikes with algorithm 2 and also by removing the average value.

In case 2., someone can't simply swap using B1 and B2 without doing additional processing or analysis. In case 1, one can simply replace B with A and perform only basic checks to compare the results obtained with the revised version against those previously used.

We also want to express

3. Dataset B is identical to dataset A in the sense that it is an exact copy of it).

4. Dataset B is almost identical to dataset A in the sense that B is identical to A, A has information appended daily, and B may be only updated once per week.

[rweigel]

Perhaps follow resampling method

isVersion has subtypes.

[elsadvd]

It seems to me that Datacite Metadata could be used to express this. You can have a look here https://support.datacite.org/docs/connecting-to-works
In particular, the following relationTypes: IsVariantFormOf, IsDerivedFrom, isIdenticalTo

[elsadvd]

Here is a list of properties for describing relations, with their definitions so that you can see which ones are most appropriate for your needs:

• dcterms:isVersionOf

definition: A related resource of which the described resource is a version, edition, or adaptation.
comment : Changes in version imply substantive changes in content rather than differences in format. This property is intended to be used with non-literal values. This property is an inverse property of Has Version.

• IVOA relationship types IsNewVersionOf

definition: The referenced resource has been developed on, and this resource should be used in preference (inverse of IsPreviousVersionOf)

• IVOA relationship types IsDerivedFrom

definition: This resource was produced using data from the referenced resource (inverse of IsSourceOf)

• IVOA relationship types IsIdenticalTo

definition: This resource is identical to the referenced resource, for use when there is a need to register two separate instances of the same resource. In a VO context, this will typcially indicate mirrors operated by different publishers (mirrors with identical publishers use interface/mirrorURL elements).

• prov:wasDerivedFrom

definition: A derivation is a transformation of an entity into another, an update of an entity resulting in a new one, or the construction of a new entity based on a pre-existing entity.

I have put aside the Datacite properties as they are not currently described in RDF.