Issue #3 in Mini-RFC

[golnazads]

Original explanation
Periodically re-validate all citing and cited identifiers (records), since bibcodes may change or even disappear over time. In particular:
Validate the citing identifier (right now a bibcode, but this could be a number of other PIDs, see below); if no known identifier is found, throw an exception
Validate the cited identifier for resolved records; if a bibcode (or future record id) associated with a resolved record is no longer in the database, attempt to re-resolve it

After discussion notes
For citing/cited

• If processing arXiv reference, if arXiv id is given, and matched, save that to the db, and that needs to happen at the level of reference service
• If doi is provided in reference string, and matched, save doi in the db, instead of the bibcode, if it matches with solr record
  Ascl identifier works the same as arXiv
• Add identifier to the returned value, but keep the bibcode
• If no arXiv/doi/Ascl do not return the identifier field
• For citing, bibcode/filename are provided, get arXiv/doi/Ascl if available from solr and save, otherwise use bibcode

[aaccomazzi]

@ehenneken please review and provide feedback

[ehenneken]

@aaccomazzi @golnazads I think this needs a wider discussion. We have multiple services that generate relationships where at least the source identifier is a bibcode. Do we require every individual service to build in logic to deal with changed/deleted bibcodes? independent of how to deal with changed/deleted bibcodes, IMHO logic to deal with matched preprints does not belong in the reference pipeline. That is a task for a downstream process. The topic related to this issue will be discussed in the wider discussion Future of managing references & citations, to be held soon.