Cataloguing rules for references in HEP records

General rules

• subfields of 999C5
  • $9: CURATOR
  • $0: Inspire recid
  • $a: DOI
  • $h: authors
  • $i: ISBN
  • $m: miscellaneous
  • $o: ref nr
  • $p: publisher
  • $r: arxiv and report nr
  • $s: journal pubnote
  • $t: title
  • $u: URL
  • $y: year

• unique identifiers
  • $r
  • $s journal ref
    • 'short journal name,volume,first page'
    • 'coden,volume, first page'
  • $a
  • $i
  • $0: 99999

The highest priority is to generate unique identifiers, esp $r and $s which were the only fields used in Spires. These subfields are used to generate links from the references to the corresponding HEP records. One identifier would in principle be sufficient, but since our metadata may not be complete it is preferable to have e.g. $r and $s if available.

• $o ref nr
  • ref nrs are now displayed to facilitate finding corresponding refs in bibedit and display
  • if one ref nr contains several references these will be split into several 999C5 fields with same $o
    • only one $o will be displayed
    • refextract sometimes splits up references by mistake, so to check whether refs with same $o should really be separate

Basic curation

• open record in bibedit
  • bibedit link is shown in body of ticket
• extract refs
  • click on box in upper left corner
  • if no refs already in record
  • if refs were extracted by older version of refextract
  • decide whether to overwrite existing refs
    • fields with $9Curator won't be touched, will produce duplicate refs
• click on preview
  • lens in upper right corner
  • shows which refs can be linked to HEP
• check refs with no links
  • if necessary open pdf via icon on top of page
  • identifier correctly recognized but no link: probably no corresponding record in HEP, nothing to be done
  • no identifier: check whether $m contains information that can serve as unique identifier
    • put into right field, possibly by inserting $$r (or s etc) in front, and curate content
  • do not clean up other fields
    • remove subfields if refextract has messed them up too much
• submit changes
  • preview will open automatically
  • quick check whether everything looks ok
  • confirm if ok, cancel if more changes need to be done

The basic task is to generate $r and $s fields, and if possible the other identifiers if time permits.

Extra curation

Anyone who can afford a bit more time to do a somewhat more thorough curation may consider the following:

• books

Known problems

• erroneous splitting up of references
  • often caused by single letters misinterpreted as author initials

-- AnnetteHoltkamp - 26-Mar-2012