DCAT-US writer: Distribution -> Access/Download URL, Media Type

[jwaspin]

distribution

mdJSON source: tbd

• [ ] accessURL
• [ ] downloadURL
• [ ] mediaType

Field Name	DCAT Name	Condition	mdJson Source
Distribution	dcat:distribution	if exists resourceDistribution[any] and if exists resourceDistribution.distributor[any].transferOption[any].onlineOption[any].uri for each resourceDistribution[0, n] where exists resourceDistribution.distributor.transferOption.onlineOption.uri then {description, accessURL, downloadURL, mediaType, title}
- Description	dcat:distribution.description	exists	resourceDistribution.description
- AccessURL	dcat:distribution.accessURL	if citation.onlineResources[first occurence].uri [path ends in ".html"] [required if applicable]	resourceDistribution.distributor.transferOption.onlineOption.uri
- DownloadURL	dcat.distribution.downloadURL	if citation.onlineResources[first occurence].uri [path does not end in ".html"] [required if applicable]	resourceDistribution.distributor.transferOption.onlineOption.uri
- MediaType	dcat:distribution.mediaType	[add codelist of "dataFormat"] transferOption.distributionFormat.formatSpecification.title [dataFormat] [dataFormat should conform to: https://www.iana.org/assignments/media-types/media-types.xhtml]
- Title	dcat:distribution.title	exists	resourceDistribution.distributor.transferOption.onlineOption.name

[dwalt]

The code is apparently mapping to distribution.distributor[ ].transferOption[ ].distributionFormat.formatSpecification.title. This works, sort of. However, it really is just a citation title, an uncontrolled field that will not necessarily conform to the DCAT media-type formats listing. Citation title doesn't seem an appropriate use for format name. It is a result of an astonishing absence of a distribution format code in ISO. Even CSDGM had at least a "recommended" list.

We can either accept that and users enter whatever they want and we will call it media type even if they entered their dog's name, or we can consider an extended ADIwg codelist to control types. IMO, this is not just an issue for DCAT, but for normalizing format types in our data documentation. This approach would be specific to mdJSON, therefore a potential ISO reader and probably CSDGM would still have to map to format spec.