What happens in OA Monitor when an open access file is superseded?

[tomwrobel]

The following question was raised by @lintuula

It relates to a binary file being superseded (i.e. replaced by a better version). An obvious example would be an Accepted Manuscript being replaced by a Version of Record. In such instances, it's unclear what would happen to the compliance status in certain circumstances.

Scenario, w. Hypothetical dates:

• On January 1, 2020 a work is accepted by a journal for publication in its February edition, due 2020-02-01.
• An object is deposited on January 2, 2020 with a binary file that is the Accepted Manuscript (AM) with a record embargo of Feb 1, 2020
• The AM has no embargo after the record embargo ends
• The journal article is published on Feb 1, 2020
• The object is published in ORA Public with the AM binary file on Feb 1, 2020
• The object is moved back to Symp with record and file publication metadata set for the AM binary file (file_made_available_date of 2020-02-01, file_public_url with the correct URL). All values are within 3 months of acceptance, publication,etc.
• Presumably the record is marked in OA monitor as Live and with a compliant binary file?
• Six months pass
• The author wishes to deposit the VoR on August 1, 2020
• Publisher requirements state that the AM and VoR cannot be published side by side
• The VoR can be made available now, and is published, it has file_made_available_date of 2020-08-01 and a file_public_url
• The AM is withdrawn and now has no file_made_available_date or file_public_url
• The record is updated on ORA-SYMP as published with a record publication date of 2020-02-01 and with one public binary file, the VoR, with a file_made_available_date of 2020-08-01 (this is not within 3 months of publication or acceptance), the previously published AM has no file_made_available_date or file_public_url

We can change the dates if I've got the important margins wrong, but would this chain of events lead to the object being marked as non-compliant?

This is not an issue in ORA3 because the value is reported from the Access database, which stores the file_date_made_available value on the object level, not the file level: it means the date "a" file was made available, not any given file.

[tomwrobel]

@jjpartridge bump

[mrdsaunders]

Thanks @lintuula.

file_made_available_date is not used as part of the OAM compliance calculation. (record_publication_date is indirectly used, since this value is used in the calculation of repository-status.)

This is therefore not an OAM issue. However it is an audit issue, since in the above scenario we would be unable to robustly demonstrate (or monitor) that a compliant version was made available in time.

We will discuss with the REF team. One possible outcome of that discussion down the line is that we propose a means of reporting on the date that a compliant version was first made available, which would presumably require a data model/process change for ORA.