Although I have no experience in using CKAN or DCAT, it seems that DCAT is not capable of expressing relations among datasets.
You might want to look into VoID  for expressing dataset relationships. Is CKAN capable of using terms from various vocabularies?
Regarding use of CKAN, you could try to find if non-GLAM institutes use subcollections too. Perhaps http://data.overheid.nl are able to help?
From: Code for Libraries [mailto:[log in to unmask]] On Behalf Of Johan Oomen NISV
Sent: donderdag 7 juli 2016 13:00
To: [log in to unmask]
Subject: Re: [CODE4LIB] Describe sub-collections in DCAT - advice very much appreciated
Many thanks for your reply.
Europeana has indeed harvested a (very small) part of datasets available in the Netherlands via different channels (aggregators and several projects). But what Europeana harvests are the datasets with metadata on record level, modelled in EDM (Europeana Data Model), *not* metadata about the datasets itself.
What we intend to do with our CKAN implementation is twofold:
(1) get a better idea of all (open) datasets in the Netherlands, how they are published (e.g. as Linked Data), to facilitate the transformation to a more distributed model in the Netherlands.
(2) and in order to do so register these datasets and model the metadata on dataset level to DCAT.
This is not something Europeana can provide for us, although our work will be valuable for Europeana on the long term.
Hope this clarifies matters,
> Op 6 jul. 2016 om 16:46 heeft Ethan Gruber <[log in to unmask]> het volgende geschreven:
> Sorry, to be a little more constructive:
> If you can describe the difference between Europeana's functionality
> now and your vision for your CKAN implementation, that would be
> helpful for providing advice.
>> On Wed, Jul 6, 2016 at 10:36 AM, Ethan Gruber <[log in to unmask]> wrote:
>> Are these GLAMs also putting cultural heritage data into Europeana?
>> You can already filter by country (that holds the work) in
>> Europeana.There are
>> 6 million objects from the Netherlands. Your energy might be better
>> spent either harvesting Dutch material back out of Europeana into a
>> separate Netherland-only interface or by focusing on integrating
>> smaller institutions into Europeana via OAI-PMH.
>> In fact, your own material are in Europeana:
>> On Tue, Jul 5, 2016 at 12:19 PM, Johan Oomen
>> <[log in to unmask]>
>>> Good afternoon,
>>> In the Netherlands, we’re working on overhauling our current
>>> (OAI-PMH) aggregation infrastructure towards a more distributed
>>> model. The aim is to create a comprehensive collection of digitised
>>> cultural heritage objects held by GLAMs across the country. A major
>>> component of the new infrastructure is a register with collections.
>>> We are using CKAN as the data management system for these collections.
>>> We are currently installing and configuring CKAN, and use DCAT for
>>> describing datasets. We are interested in seeing other examples of
>>> registries that describes digital heritage collections using the
>>> CKAN software. One of the challenges we encounter is describing
>>> multi level datasets like collection and sub-collections in the
>>> context of DCAT. An example is a data provider in the Netherlands
>>> that provides an aggregated oral history dataset for target audience
>>> ‘oral history’. We registered this aggregated dataset, but we also
>>> want to register individual collections for participating
>>> organisations. Therefore, the aggregated dataset is divided into
>>> parts using xpath, xslt, etc.. Now we want to explicitly mark the dataset parts as being a sub-dataset and vice versa.
>>> Question to this community, do you have implementations that use a
>>> CKAN based registry for digital heritage collections, have you also
>>> dealt with this issue to describe sub-collections in DCAT? How did you manage this?
>>> Your help is much appreciated,
>>> Best wishes,
>>> Johan Oomen
>>> Netherlands Institute for Sound and Vision @johanoomen