Difference between revisions of "DCAT LC comments"

From Government Linked Data (GLD) Working Group Wiki
Jump to: navigation, search
(Add table with all DCAT LC feedback. Status / response/ ack columns still empty.)
(Expand Ed Simon's comments with details)
Line 102: Line 102:
 
|-
 
|-
 
|| [http://lists.w3.org/Archives/Public/public-gld-comments/2013Apr/0068.html Ed Simons] ||
 
|| [http://lists.w3.org/Archives/Public/public-gld-comments/2013Apr/0068.html Ed Simons] ||
Seven comments, three suggestions
+
Issues
 +
* How to express dataset relationships such as "derived from"
 +
* Time-related properties don't allow stating what agent caused the event
 +
* Publisher is not the only dataset-agent relationship; creator and maintainer are important
 +
* Catalog language is superfluous and ambiguous
 +
* No license attached to each individual dataset metadata record
 +
* Link from CatalogRecord to Catalog not necessary; link to Dataset sufficient
 +
* Distribution types shouldn't be implicit
 +
 
 +
Suggestions
 +
* Add classes that represent connections between DS-DS, DS-Agent, Cat-Agent
 +
* Add Person and Organisation class
 +
* Add classes for each type of distribution
 
||
 
||
 
||
 
||

Revision as of 10:04, 11 April 2013

Comments and draft responses for the DCAT Last Call ending 2013-04-08.

Disposition of comments
Comment Summary Status GLD response Acknowledgement
Aaron Couch
  • List all properties in ToC
  • 2 editorial comments
Jeni Tennison

Use dct:rights instead of dct:license

Jeni Tennison

Consider adding dct:provenance or prov:has_provenance

Paul Groth
  • Consider adding dct:provenance or prov:has_provenance
  • Consider referencing PROV
Stuart Harrison

Can DCAT describe APIs?

Vasily Bunakov

Add a dcat:interface property that points to a statement or manifest about technical means of accessing the dataset

Marios Meimaris

Add a property that links a distribution to an external concept that describes its access rules and methods

Alasdair Gray
  • Move textual definitions into the tables for each property
  • 1 editorial comment
Phil Archer

Don't say that 2005-01-01 should be used for "unknown date in 2005"

Stasinos Konstantopolous

Don't say that 2005-01-01 should be used for "unknown date in 2005"

Alasdair Gray
  • Rename dcat:dataset and dcat:distribution to dcat:hasDataset and dcat:hasDistribution
  • (Note resulting long discussion thread)
Bill Roberts
  • Add something for Granularity
  • Clarify how to use accrualPeriodicity
Andrea Perego

Add properties for dataset versioning, along the lines of ADMS

Christopher Gutteridge

Add Contact and Corrections properties

Andrea Perego

What exactly is the scope of DCAT? Datasets or any kind of resource? Consider extending the scope.

Ed Simons

Issues

  • How to express dataset relationships such as "derived from"
  • Time-related properties don't allow stating what agent caused the event
  • Publisher is not the only dataset-agent relationship; creator and maintainer are important
  • Catalog language is superfluous and ambiguous
  • No license attached to each individual dataset metadata record
  • Link from CatalogRecord to Catalog not necessary; link to Dataset sufficient
  • Distribution types shouldn't be implicit

Suggestions

  • Add classes that represent connections between DS-DS, DS-Agent, Cat-Agent
  • Add Person and Organisation class
  • Add classes for each type of distribution
Makx Dekkers

The mental model of catalogs + datasets, and whether they can be used independently, is not really clear

Makx Dekkers

Is it implied that a dcat:Dataset must have at least one distribution on the Web

Makx Dekkers

What is the "single source" that the dcat:Dataset definition talks about?

Makx Dekkers

Spec presents Concept and ConceptScheme in a single box despite being separate classes

Makx Dekkers

Not clear what distinguishes a dataset from rdfs:Resource

Makx Dekkers

Range of dct:language should be dct:LinguisticSystem

Makx Dekkers

DCAT should not place restrictions on use of SKOS