Headstart icon indicating copy to clipboard operation
Headstart copied to clipboard

CORE: Implement parameters from original specification

Open pkraker opened this issue 7 years ago • 9 comments

The CORE connector implements different parameters than the original specification. They were:

  • from: publication date lower bound in the form YYYY-MM-DD
  • to: publication date upper bound in the form YYYY-MM-DD
  • article_types: in the form of an array of identifiers of article types
  • langs: in the form of an array of language identifiers (["all"] for all languages)
  • sorting: can be one of "most-relevant" and "most-recent"

From my point of view, there is no harm in keeping the additional parameters (the key param is even mandatory in the current implementation), but these five parameters should also be implemented to enable our usual query refinement options.

pkraker avatar Aug 22 '18 11:08 pkraker

I'm currently working on this, but this requires a few additions to rcoreoa. I will open a PR with additional query params in rcoreoa soon.

The additional params 'repositoryDocument.metadataUpdated' and 'language.name' of the CORE elasticsearch API are documented here

chreman avatar Aug 22 '18 11:08 chreman

It also appears that some functionalities will have to diverge from our current ones - CORE only provides article publication date with a yearly resolution. The parameter 'repositoryDocument.metadataUpdated' may be misleading because it refers to the date when articles were added to CORE, not when they were published. Furthermore, 'article_types' is not a usable field, and sorting is only available via a relevance score. I will contact CORE for clarification.

chreman avatar Aug 22 '18 12:08 chreman

Thanks! Just a quick comment regarding publication date: I have seen the field datePublished, which seems to carry the publication date in the best resolution that they have. This would be similar to how publication date is reported in BASE and should do for our use case.

pkraker avatar Aug 22 '18 13:08 pkraker

Are you referring to datePublished as returned by rcoreoa? I think this is the value we get in the returned metadata, but not what CORE allows us to search in their API.

chreman avatar Aug 22 '18 20:08 chreman

@sckott I attempted an implementation of a language search param in rcoreoa, please see PR. Unfortunately this does not have any effect as some of my tests with English/en, French/fr, or German/de showed - but perhaps I'm missing something in the rcoreoa - thanks for having a look!

chreman avatar Aug 22 '18 21:08 chreman

thanks

sckott avatar Aug 22 '18 21:08 sckott

@chreman did feedback

just in case: note that the core_journals function is broken right now i think becaus somethin's broken on their end https://github.com/ropensci/rcoreoa/issues/17

sckott avatar Aug 22 '18 21:08 sckott

erroneously closed due to number typo

chreman avatar Jun 19 '19 21:06 chreman

@chreman Have you tried this recently? It should all be working on the rcoreoa side of things

sckott avatar Oct 01 '19 16:10 sckott