search

CSIRO-enviro-informatics / asgs-dataset

GNU General Public License v3.0
0 stars 1 forks source link

Create a comprehensive metadata record for ASGS data #8

Open dr-shorthair opened 5 years ago

dr-shorthair commented 5 years ago

https://github.com/CSIRO-enviro-informatics/asgs-dataset/blob/master/asgs_dataset/view/dcat.ttl was created as a basic metadata record. It should be (a) updated to match DCAT-2 best practice (b) serialised in JSON (c) embedded in the asgs-dataset and landing page header

Supersedes #3

nicholascar commented 5 years ago

@dr-shorthair note that the pyLDAPI can serve up conneg for the landing page so while it's a great idea to embed the DCAT2 record as JSON-LD in the HTML landing page, please can the same content in RDF be available at curl -H 'Accept: text/turtle' http://linked.data.gov.au/dataset/asgs2016 etc.? This makes AGLDWG cataloguing even easier than parsing the JSON-LD and is the correct Linked Data thing to do.

The DCAT content can all be taken from 1 file in pyLDAPI, for the JSON-LD-in-HTML and RDF delivery formats.

ashleysommer commented 5 years ago

@nicholascar Yes, I had already planned to make the same metadata available as both embedded json-ld and in the RDF representation of the dataset.

dr-shorthair commented 5 years ago

OK - I see that request is currently getting the 'DCAT' record that you created.

Please inspect and critique the updated version I've now made - https://github.com/CSIRO-enviro-informatics/asgs-dataset/blob/master/asgs_dataset/view/dcat.ttl

I distinguished three resources in this record:

  1. the original ASGS dataset, packaged by ABS Geospatial Solutions and delivered as a Geopackage - http://data.gov.au/dataset/asgs-2016-edition-boundaries
  2. the linked-data version of the dataset, structured according to asgs-ont - http://linked.data.gov.au/dataset/asgs2016
  3. the service endpoint at https://asgsld.net/2016/

There is a question as to whether delivering the metadata record would meet expectations, when de-referencing the dataset URI. Of course this is the issue addressed by conneg-by-profile, but should probably also be reflected in the 'alternates' view.

nicholascar commented 5 years ago

should probably also be reflected in the 'alternates' view.

I'll check over all the 'alternates' views shortly since I'll be looking for evidence of implementation for Conneg by Profile & PROF. I'll push a change to pyLDAPI that will do most of the work but will work with @ashleysommer and others to ensure the change is non-breaking for all the LocI implementations.