HoldingPen: Test matcher

[ksachs]

to compare the result of the matcher with BibMatch with higher statistics we need to align the DESY lists (by feed / xml file) and the HP result.

For a few files it's OK to filter by publisher which is (to some extend) searchable.
Usually the publisher is in e.g. abstracts.source, but not all records have an abstract.
So this is not reliable. In addition there can be multiple feeds (different days or journals) submitted for the day, i.e. a cataloger has to look at several feeds to check whether there is a match.

Searching for the journal-title in the HP is too cumbersome unless there are facets.

It is not possible to search for the date of the harvest, which is OK as long as we clean the HP daily.

If BibMatch finds a match and the record is not halted in the holdingpen for matching one has to search manually record by record. If it is

• found by DOI or arXiv: we would assume the matcher finds it too and don't search for the record in the holdingpen.
• otherwise one has to search via DOI, e.g. metadata.dois.value.raw:"10.1103/PhysRevD.97.115023"
  if the record is halted due to conflicts: means there was an automatic match
  if the record is halted for selection: means there was no match
  but what if the record can not be found in the holdingpen (as in the example above?)

I doubt the current possibilities are enough to do thorough testing.
However, it is not worthwhile do develop anything fancy.

Can someone can come up with a solution for the holdingpen?
Or we switch to processing via holdingpen (after the open issues are fixed) and develop some cross-check based on the DESY workflow.

[ksachs]

When processing via holdingpen incl. updates to INSPIRE is active we will do a cross-check based on the DESY workflow.