How to update article metadata

[jvwong]

The purpose of the article update module is to add and/or refresh metadata for an existing paper. This must be done carefully, taking into account:

• intentions of the user
  • which (version) of a paper they are referring to
  • what type, subtype of paper they are referring to (preprint, article)
• type of paper information supplied by the user
  • normally a title
• the way different metadata sources handle paper info
  • availability of identifiers (PMID, DOI).

Would an overwrite delete all the previous ‘pubmed’ data, or would it be a merge? Do you see a use case for merging? Deletion would be simpler.

Another alternative is to store the original, raw data from each source so you could always recover or merge in future, i.e.

• pubmed stores the processed data the app expects
• rawPubmed stores the original, raw PubMed data
• rawCrossRef stores the original, raw CrossRef data

I think straightforward deletion is the simplest approach, but let me know if you see a use case for the other approaches.

This (CRON updates/disambiguation of papers) deserves a separate issue, apart from CrossRef (todo).

One latent bug that has emerged is: PubMed will now index preprints from bioRxiv/medRxiv authored by those NIH-funded, so, e.g. someone could add a bioRxiv paper picked up by PubMed and if it was published somewhere with the same title (i.e. author input) it would get overwritten.
You can always put in the exact PMID or DOI but I suspect 90% of people have not idea what those are.

Originally posted by @jvwong in https://github.com/PathwayCommons/factoid/issues/961#issuecomment-1708773205

[jvwong]

Below are the possible cases for an article that needs to be updated (i.e. not initiated, not empty and not stale).

Table 1. Article update cases (in order of preference)	paperId (provided)	PMID	DOI	Status	Action
Y	Y	Y/N	In PubMed	GET - via PMID
Y	N	Y	Only in CrossRef	GET - via DOI
Y	N	N	Not found	SEARCH - query with paperId (text)