Fix issues with current triplification pattern

amoeba commented 2 years ago

Last week we worked through some example triples and found a few issues and areas for improvement. See https://hpad.dataone.org/9cvSZ4l0RaK-FHjlOcKmPw for a list of the hackpads we worked through. We pointed out issues directly in the hackpad, and an abridged list is:

[ ] Missing PROV in at least one case
[x] Award structure doesn't match SOSO
[x] Add all parties from metadata not just creator
[x] Parties aren't following SOSO pattern
[x] Missing checksum algorithm triple
[ ] Organization identifiers via ROR
[x] variableMeasured triples could be more complete
[ ] Normalize people and org names
[x] Updated funding pattern from SOSO. CreativeWork <- funding -> MonetaryGrant

See the linked hackpad for more info around each one. Most of these are quick fixes and can go in before our first release.

amoeba commented 2 years ago

Also revisit this paragraph to make sure all of this is captured:

I did a quick read through, @amoeba -- looking good. While there were a lot of particulars that we can pick up on later, a couple high-level points. For provenance, I think we should follow the DataONE provenance model as expressed in the SOSO provenance guidance. This will be quite compatible with Carl's approach, which is already pretty isomorphic with ProvONE albeit lacking some specific things. Second, when we have multiple values for a field, like for identifier with have PID, SID, and alternative Identifier fields, we can and should include them both, and follow SOSO guidance on indicating precedence and using sameAs in reasonable ways. For your example, it would be good to use an example that is pretty comprehensive and has both annotations and detailed provenance information to show how those would work. Finally, we should discuss the Dataset/Metadata/Entity/Package distinctions and how we will map those -- I was a little confused by your initial text around those. Nice work all around!

-- From https://github.com/DataONEorg/slinky/pull/21#issuecomment-819130683

amoeba commented 2 years ago

Also bringing this in from our weekly call Re: EML units,

EML Units in slinky

Should bring the unit dictionary into Slinky as OWL and use that in triples (to avoid using string literals

Remember we made some changes to the units in 2.2.0, should upgrade any deprecated units we find in 2.2.0 docs to the new code

amoeba commented 2 years ago

Made some good progress on this today and found a few more things that could use some work. I'll add those to the checklist above as I go.

amoeba commented 2 years ago

Noting here that https://schema.org/funding is live.

amoeba commented 2 years ago

Added support for schema:funding in https://github.com/DataONEorg/slinky/commit/7aa9d9487315988db12732aeac732032f21431a2 by mapping EML's award structure. We now get triples out like:

schema:funding <https://www.nsf.gov/awardsearch/showAward?AWD_ID=1417345> ;

<https://www.nsf.gov/awardsearch/showAward?AWD_ID=1417345>
    a schema:MonetaryGrant ;
    schema:funder <https://doi.org/10.13039/100000001> ;
    schema:identifier "1417345" ;
    schema:name "COLLABORATIVE RESEARCH:   Ice Regime Shifts of Arctic Lakes Drive Interactions and Feedbacks with Permafrost and Climate" ;
    schema:url "https://www.nsf.gov/awardsearch/showAward?AWD_ID=1417345" .

<https://doi.org/10.13039/100000001>
    a schema:Organization ;
    schema:identifier "https://doi.org/10.13039/100000001" ;
    schema:name "National Science Foundation" .

DataONEorg / slinky

Fix issues with current triplification pattern #64