Dataset 0.5 DRAFT Comments

[susheel]

Comments on the 0.5-DRAFT schema just mapping this to our HDR UK schema and DataCite

Minimum:

• Should we have some notion of the publisher of the dataset - Crosswalk to DataCite and HDR UK
• Year of publication - Crosswalk to DataCite and HDR UK
• Creator should be minimum - Crosswalk to DataCite and HDR UK

Recommended:

• GA4GH DUO code for accessRights - Link to ELIXIR and GA4GH
• dataController (as per GDPR) if different from publisher / creator
• jurisdiction/geolocation - GDPR

[AlasdairGray]

These are some really important points, particularly when thinking about the Open AIRE Research Graph use case.

[AlasdairGray]

In terms of the use cases we are trying to support, I think that your suggestions are one level too high, i.e. the properties you suggest for minimal should be recommended and those for recommended should be optional.

[egonw]

My 2 cents...

Generally, I would support encouraging DataCite support, but that alignment should be at profile level first, I think: "The Bioschemas Dataset profile is aligned with ..."

Other points:

• codes for accessRight is for interoperability. that would mean making the type a TERM
• dataController: this I think do not belong in the profile: the controller is not linked to the data: anyone reading the Bioschemas is a controller too (new process in GDPR terms), so that makes ZERO sense
• jurisdictions is similar: the license is not limited to the origin and for many licenses the jurisdiction is meaningless (like CCZero). Also, GDPR also applies if data comes from a "non-GDPR" jurisdiction, because GDPR is focussed on the processing.

Now, IANAL, but this insight follows from discussions with legal experts. Happy to learn more about GDPR :)

[AlasdairGray]

Based on the community discussions, here are the responses to your points. Note that we aim to keep the Dataset profile aligned with the Google Dataset Profile unless there are compelling domain specific consuming use cases to deviate. We also strive to keep the minimal set of properties to about 6.

Note that the Schema.org vocabulary is explicitly aligned with DCAT, and there are straightforward correspondences of many terms to DataCite.

Comments on the 0.5-DRAFT schema just mapping this to our HDR UK schema and DataCite

Minimum:

• Should we have some notion of the publisher of the dataset - Crosswalk to DataCite and HDR UK

publisher property already included, increased to recommended marginality

• Year of publication - Crosswalk to DataCite and HDR UK

In response to this issue, we have promoted datePublished to recommended. Note that the Google profile makes no recommendations about dates associated with a Dataset.

• Creator should be minimum - Crosswalk to DataCite and HDR UK

creator has been left as a recommended property

Recommended:

• GA4GH DUO code for accessRights - Link to ELIXIR and GA4GH
• dataController (as per GDPR) if different from publisher / creator
• jurisdiction/geolocation - GDPR

These are specific to the HDR publication use case but at this time there is no clear markup consuming use case. For now we will not make any changes but will review if a specific consuming use case arises which could include have a specialised version of the profile for datasets involving personal identifying data.