ACTION-267: Draft reply to Clemens

Draft response to Clemens Portele –
=============

Dear Clemens –

Thank you for your thoughtful comments on the proposed revision of the W3C Dataset Catalog Vocabulary. As you note, the addition of classes for data-services does appear to extend the scope of DCAT. However, we believe this extension to the vocabulary is necessary.

Many current data catalogs already list data services. This was reflected in a number of submitted use-cases [1] which requested better guidance in DCAT for how to do it. As you note, some had used the `dcat:Distribution` class, whose definition in DCAT 2014 appeared to allow for this. However, the patterns for cataloguing data services were unclear and catalogs are inconsistent.

After considerable discussion, which is reflected in several GitHub threads [2], the DXWG chose to resolve the issue by (1) adding a new class dcat:DataService with some sub-classes, and (2) clarifying (limiting) the scope of `dcat:Distribution`.

A `dcat:Distribution` is now intended to be used to describe a specific representation or serialization of a dataset.

A `dcat:DataService` is intended to be used to describe a particular endpoint that delivers data or data processing functions.

A data-service will usually accept parameters for configuration or queries, etc, so conceptually a large number of implied datasets are available from a service, most of which would not be described explicitly prior to request. Furthermore, while some data-download-services deliver subsets of one or a small number of datasets (which might be described in separate entries in a catalog), some data-processing-services are not tied to any particular dataset (e.g. the pedo-transfer-function service deployed in the OGC Soil IE which could take input from several data services). For the latter case in particular it is not clear how a service not tied to a dataset could be included in a DCAT Catalog using the `dcat:Distribution` class, since it is not a first-class member of a catalog.

To provide an overarching framework, we also added a (implicitly abstract) super-class `dcat:Resource` for all resources in a DCAT catalog. The new DCAT vocabulary is unlikely to define any further sub-classes, but external profiles might add classes for many other types that might be catalogued, such as samples, instruments, facilities, projects, … We believe that it would be helpful to provide a common semantic superstructure for this, though things beyond datasets and data-services are out of scope of the revision of DCAT.

We hope this clarifies the thinking of the DXWG and reassures you that this extension was not done casually.

[1] https://www.w3.org/TR/dcat-ucr/#ID6 , https://www.w3.org/TR/dcat-ucr/#ID18
[2] https://github.com/w3c/dxwg/issues?q=is%3Aissue+is%3Aclosed+label%3Aservice

Response sent to public comments list 2019-01-10