RE: reviewing the BP doc

*  I don't think that a Dataset is an abstract thing. But I agree that distributions of a Dataset (DCAT definition) are instances of the same Dataset source.  And this is one of the possible relations between Datasets.

In my perspective, a Dataset *is* an abstract thing. It only physically exists in its Distributions. For example, a Dataset may not have Distributions, e.g. if a description of a Dataset is generated before the data is collected (the Dataset of tomorrow’s weather observations) or a description still exists in the Catalog after the data files have been deleted. In my mind, that means a Dataset is an abstract entity.

The description of ADMS http://www.w3.org/TR/vocab-adms/, a specification closely related to DCAT, is more explicit about this. It defines  adms:Asset, a subclass of dcat:Dataset, as: “An abstract entity that reflects the intellectual content of the asset and represents those characteristics of the asset that are independent of its physical embodiment.”

Makx.