RFC Reference matching algorithm

[kaplun]

Let's assume we have workflow "reference matching":

• [ ] if not raw reference:
  • [ ] run refextract
  • [ ] store refextract ref as raw ref of type refextract
• [ ] for every reference
  • [ ] translate reference into mini-record
  • [ ] try to match mini-record with existing one
  • [ ] if sure match
  • [ ] record the match
  • [ ] elif ambiguous match and CORE:
  • [ ] request cataloguer to check
  • [ ] else:
  • [ ] create stub-record
  • [ ] record the match

[annetteholtkamp]

What do we define as minimal metadata necessary for creating a stub record?

• Annette

On 25 Aug 2016, at 08:15, Samuele Kaplun notifications@github.com wrote:

if not raw reference run refextract for every reference translate reference into mini-record try to match mini-record with existing one if match record the match else: create stub-record record the match — You are receiving this because you are subscribed to this thread. Reply to this email directly, view it on GitHub https://github.com/inspirehep/inspire-next/issues/1486, or mute the thread https://github.com/notifications/unsubscribe-auth/AM1-O-M2_G0uwkWGn3-P11m7FnViD4KFks5qjU8ogaJpZM4JszkN.

[kaplun]

I'd say:

• arXiv ID (because we can fetch the matadata on the fly from arXiv
• DOI (same reason from Crossref)
• ISBN (because we can fetch the book on the fly)
• journal, volume/issue, (year), pages (because it allows to unambiguously match it)
• reportnumber + title + author

[StellaCh]

@jacquerie shall we close this?

[jacquerie]

@jacquerie shall we close this?

We shall.