W3C home > Mailing lists > Public > public-dxwg-wg@w3.org > June 2018

[dxwg] Catalogues in which dataset is a bag of files

From: Simon Cox via GitHub <sysbot+gh@w3.org>
Date: Tue, 19 Jun 2018 20:36:08 +0000
To: public-dxwg-wg@w3.org
Message-ID: <issues.opened-333829129-1529440567-sysbot+gh@w3.org>
dr-shorthair has just created a new issue for https://github.com/w3c/dxwg:

== Catalogues in which dataset is a bag of files ==
### Catalogues in which dataset is a bag of files
Status: Proposed

Identifier: ID53 (Proposed)

Creator: Simon Cox

Deliverable(s): DCAT

## Tags
dcat, dcat:Dataset

## Stakeholders
Operators of legacy catalogues, consumers of legacy catalogues

## Problem statement
In many legacy catalogues and repositories (e.g. CKAN), ‘datasets’ are ‘just a bag of files’. There is no distinction made between part/whole, distribution (representation), and other kinds of relationship (e.g. documentation, schema, supporting documents). 

While the precision we provide in DCAT is valuable in terms of semantics, it is often difficult to implement on these legacy systems. In particular, we only recommend one property to link from `dcat:Dataset` to another artefact, i.e. `dcat:distribution` which is designed to link from a dataset to a _representation_ of the whole dataset. Guidance is required for linking to resources which are deemed to be elements of a dataset where the nature of the relationship is unspecified. 

## Existing approaches
People mostly use the `dcat:distribution` relationship for _all_ links to dataset elements. This is strictly incorrect in many cases where `dct:hasPart`, `dct:conformsTo`, `dct:requires`, `dct:references`, or another relationship, would be better . 

## Links
''Optional link list to documents and projects this use case refers to''

## Requirements
1. add a recommendation ('should' statement) to use `dct:hasPart` to link from a `dcat:Dataset` to a file or other resource that is part of the dataset package, _where the nature of the relationship is unknown_
2. add the axiom `dcat:distribution rdfs:subPropertyOf dct:hasPart .` so that the recommendation is _consistent with all potential relationships_ between datasets and their elements

Note: other relevant relationship predicates `dct:conformsTo`, `dct:requires`, `dct:references` are already sub-properties of `dct:hasPart`. 

## Related use cases
ID32 - Relationships between datasets

## Comments
''Optional section for editorial comments, suggestion and their interactive resolution''


---


Please view or discuss this issue at https://github.com/w3c/dxwg/issues/256 using your GitHub account
Received on Tuesday, 19 June 2018 20:36:14 UTC

This archive was generated by hypermail 2.3.1 : Wednesday, 30 October 2019 00:15:43 UTC