Consider a JSON-LD Guide for Darwin Core

[tucotuco]

Just an idea awaiting demand at this point.

[baskaufs]

I think this may be a good idea moving forward as we consider how serializations might be interconverted to making linking across standards easier.

I think that it would be good to create this together with a revision of the RDF Guide since one of the points of JSON-LD is that it can be interpreted as RDF. The RDF Guide was constructed with a particular outlook on machine-readable data. The outlook on machine readable data has evolved somewhat with the development of practical implementations like Wikidata, schema.org, and JSON-LD, and the RDF guide should probably reflect those developments.

[csbrown-noaa]

I am interested in exploring this. Specifically, I have a use case that involves combining non-DwC info with DwC info. The existing DwCA framework does not facilitate this. However, other RDF/CSV standards should empower interoperability of DwC with arbitrary RDF-compliant vocabularies. Specifically Croissant or csvw would allow referencing the tabular data within an RDF-compliant framework. The RDF-compliance would then open up the world of linked data to DwC datasets.

I would like to do some proof of concept work on converting DwCA metadata.xml into Croissant+DwC JSON-LD. Would be happy to report back on lessons learned and the ability of existing Croissant libraries to automatically parse and load the associated tabular data files.

I would appreciate some advice from the TDWG community that this isn't a total waste of time. :) Anyone have thoughts on this project?

[baskaufs]

I think there are two pieces to moving forward on this:

1. Providing use cases and proof of concept examples to suggest approaches that might work.
2. Determine whether there is enough demand for it that multiple stakeholders would actually use it to share data.

So on point 1, it would be great if you could come up with examples and situations where they might be useful. Then we might check to see how great the interest is and whether it would be worth the effort to form a task group to develop it as either part of the standard, or as a recipe for a "layer" on top of the basic bag of terms (kind of a suggested best practice without requiring going through the standards process to develop and change it).

A couple other comments. I've been interested in the possibilities of the W3C csv2rdf standards as a way to make tabular data interpretable as RDF (see https://doi.org/10.3233/SW-210443), but when I've discussed it with people in TDWG, there hasn't been a lot of interest since it hasn't gotten a lot of traction generally and there aren't many tools for using it. Rather, people seem pretty excited about Frictionless data packages. It seemed to me that a JSON-LD metadata description file that maps the columns of a tabular data file (e.g. CSV) to RDF triples could be included with the Frictionless data package to make it interpretable as RDF, but again, would anybody use it? I'm not familiar with Croissant.

The other thing I would mention is that there was considerable discussion in the Technical Architecture Group a few years ago about strategies for expressing "flattened" complex data fields in a structured format like JSON. There's a summary of the discussion at https://github.com/tdwg/tag/blob/master/meetings/2023-05-08-tag-meeting-notes.pdf . So I'd take a look at this to see what the issues were that were discussed.