search

earthref / MagIC

EarthRef's MagIC Web Application
https://earthref.org/MagIC
MIT License
8 stars 2 forks source link

Duplicate MagIC contributions #552

Open njarboe opened 2 years ago

njarboe commented 2 years ago

A list of contributions that have been found to have duplicates:

A least one contribution recently published: Rais et al, 1996 has two versions. 19405 and 18393

PINT08 fix - Riisager & Perrin 1999(19414) or Riisager, et al. 1999(17466) are duplicates. Strangely one of the Riisagers is dropped from the author list when the info is retrieved from DataCite, but there are three authors on the paper as referenced by the paper doi: 10.37570/BGSD-1999-46-07

PINT08 fix - Bucha 1968: 19420 and 17338 should be merged.

"PINT " fix - Tanguy 1975: 19433 and 19431 should be merged.

other duplicates: PINT08 fix - Brandt 2009: 19419 and 17267 should be merged. 19419 cannot be published at the moment due to BinInt problem.

christeanne commented 2 years ago

external_database_id: "PINT " typo fix

Needs to be merged:

rminnett commented 2 years ago

These are merged now:

The notebook that does the merging still needs to be checked carefully since the changes are difficult to reverse - I'll see if I can add some additional checks to make it a bit more robust.

christeanne commented 2 years ago

external_database_id: "PINT " typo fix

Needs to be merged:

njarboe commented 2 years ago

The new ones Christeanne are creating seem to be all creating new contributions instead of adding new versions to an old one.

christeanne commented 2 years ago

The new ones Christeanne are creating seem to be all creating new contributions instead of adding new versions to an old one.

The Laj and Kissel (1999) contribution that I uploaded updated the preexisting versions as expected, but all of the other contributions in the "PINT " list created an entirely new version. I have been uploading them all the same way so I'm not sure what's going on!

rminnett commented 2 years ago

It looks as though it's a bug with matching previous version histories if the new contribution has a DOI assigned using the dx.doi.org link. We allow this as a valid reference and it retrieves the metadata from Crossref, but it doesn't match the old version history if that just had the DOI instead of the link. The fix for now is to try setting the reference to either the DOI or the dx.doi.org link (you might have to try both, it just has to match the previous version history) and check to see if it assigns a new version instead of version 1 before publishing it.

For example, like this: image

Instead of like this: image

I'll deploy a fix for this and update all existing contributions that have a dx.doi.org link stored as the reference so both a dx.doi.org link or the DOI itself will work at matching the version history and store the reference as just the DOI.

rminnett commented 2 years ago

These are merged now:

christeanne commented 2 years ago

external_database_id: "PINT " typo fix

Needs to be merged: