search

DataONEorg / sem-prov-ontologies

Ontologies focused on scientific observations and scientific workflow provenance.
https://ontologies.dataone.org
17 stars 7 forks source link

Add gh-pages website for repo, MOSAiC 1.0.1, and various MOSAiC extras #107

Closed amoeba closed 3 years ago

amoeba commented 3 years ago

This PR adds a GH Pages-built website for the entire repo, see https://dataoneorg.github.io/sem-prov-ontologies/. It's set up to only build off the main branch. See #101.

This PR also adds version 1.0.1 of the MOSAiC ontology (#104) and adds some extra features: SHACL validation (#103) and correspondence checking between PANGAEA's info and our ontology (#102).

Once merged, I'll tag a MOSAiC 1.0.1 release off of the merge commit.

amoeba commented 3 years ago

Hey @mbjones, do you want to take a look here before I merge? The MOSAiC 1.0.1 ontology and related stuff is pretty straightforward but I think the new website build is worth a look. It's super barebones and could be made a lot snazzier if we like but it's a start.

mbjones commented 3 years ago

Sure, I can take a look. Thoughts on why the MOSAiC correspondence check is failing with the error:

Exception: Campaign with label 'KD-MOSAiC-1' not found in the expected set ['PS122/5', 'PS122/4', 'P5_223_MOSAiC_ACA_2020', 'PS122/1', 'PS122/2', 'PS122/3', 'AF-MOSAiC-1', 'P6_222_IceBird_MOSAiC_2020', 'AT-MOSAiC-1']

Is that expected and you're just ignoring it for now?

mbjones commented 3 years ago

Oh, and also the correspondence check is failing, which seems like we should fix.

amoeba commented 3 years ago

The only issue I saw was the lack of owl:equivalentClass and owl:equivalentProperty statements to link the new https URIs to the older http URIs. As we have already annotated datasets against the http URIs, what is the plan there to support the crosswalk?

When we decided to make the change, I checked in with @laijasmine to see what had already been annotated. IIRC it was only about four annotations in total and we updated the affected documents with the corrected IRIs. And all affected records were never publicly-readable. I felt this was good enough at the time and didn't want to go down the route of issuing the equivalences you mention. Does that sound good enough to you?

Oh, and also the correspondence check is failing, which seems like we should fix.

Yep, good catch. We'll have to coordinate who'll do the work on that (we're missing a Campaign).

I'll leave this PR open until we coordinate on the equivalence issue above.

mbjones commented 3 years ago

Your approach to the http/https equivalence seems great given that the annotated versions were all private. So, yeah, let's go with this.

N.B. The same http/https issue will likely arise for other ontologies that may be more used in the wild.

amoeba commented 3 years ago

N.B. The same http/https issue will likely arise for other ontologies that may be more used in the wild.

Right, good point. Thanks.