Withdrawn papers

[kaplun]

Withdrawn papers must:

• [ ] be tagged with "withdrawn": true in JSON
• [ ] not be listed in author profile or in any citesummary
• [ ] however be matchable when resolving references
• [ ] however be searchable in literature
• [ ] clearly marked as withdrawn when displaying it

cc: @inspirehep/inspire-content

[michamos]

When a record gets flagged as withdrawn, the following should happen automatically (from @annetteholtkamp in the recent thread on INSPIRE-DATA):

The rules we agreed on I think are:

Remove everything except the bare metadata so that these papers don’t show unnecessarily up in searches. So keep only:

• title (I usually put “Withdrawn:” in front to make it more visible)
• authors (not their affiliations)
• arXiv note and abstract
• arXiv nr and rep nrs Remove everything else Add 980:Withdrawn

The title modification can be avoided if withdrawn records are displayed differently.

Unwithdrawing the record should revert to the latest non-withdrawn revision in order to restore all metadata including enhancements.

[ksachs]

keep also

• DOI
• pubnote (773)
• hidden notes ?

What do we do with dates, pagenumbers, subject, pacs, license, hidden notes, attached files, ...? I would delete the abstract(s) too.

Would it be better to define what should be deleted:

• affiliations
• keywords
• subject?
• pacs?
• references

[michamos]

@ksachs If it is withdrawn from arXiv but there is a non-withdrawn published version, should the record really be withdrawn?

[ksachs]

Even a journal article can be withdrawn: https://inspirehep.net/record/1486153?ln=en https://inspirehep.net/record/1377762?ln=en

[ksachs]

whether or not https://inspirehep.net/record/1489216?ln=en should be flagged as withdrawn is another (content) question

[michamos]

You are right, I was thinking about withdrawn on arXiv, not about retracted published papers (which should be handled consistently).

[aw-bib]

Even a journal article can be withdrawn

Note that there are some problematic corner cases. E.g. some publishers mark articles as withdrawn if they actually made a mistake (e.g. published the article twice, registered two DOIs). I understood, that the publishers have "no other means to handle this".

So, in the interest of the authors some sort of manual check might be in order.