Service for Control of Named Entities/Discussion

From MPDLMediaWiki
Jump to navigation Jump to search

Named entities discussion[edit]

  • Input from Ingo Brüggemann (08.09.2008)

Ich hätte, um auf die Stoßrichtung meiner Frage zurückzukommen, erwartet, dass die Personenangabe nach Einführung der named entities für Personen dann schlicht '<escidoc:identifier>person:170</escidoc:identifier>' und nichts weiter lauten müsste. Denn was passiert, wenn man später das 'J.' auflösen möchte? Werden dann alle Fedora-Eingträge mit 'person:170' nochmal angefasst? Oder sieht das nur in der Ausgabe so aus?

  • Input from --Natasa 15:34, 8 September 2008 (UTC)
    • we have to keep all name variants, and still be able to relate them with the same person identifier.
    • Sometimes, person identifier would even not exist (i.e. during import) -> would mean we have to develop better procedures that match names better and run under proper QA (Quality assurance).
    • Of course, ideally we simply should link the personID to the Item resource (and not deal with first/last name) of persons.

This would require a very well established procedure of author/person management by all institutes that would use PubMan, and as much as possible automatic matching. Maybe we can come to that point of organizational matureness sometimes - For now, we provide both possibilities.(sure, there may be institutes that are very well in establishing this procedure within some institutes -> we look forward to information on such efforts and experiences).

    • It is of course not at all recommendable to change all possible Fedora entries that contain the reference to person:170 - and we do not intend to do it.
    • Name change can really be the consequence of:
      • error (false name given), or
      • real change of name (Mustermann is now Schmidt :))
    • Old publications should still be referencable and searchable for Mustermann, but also should be referencable/searchable by Schmidt. The same case would have to be for new publications.
    • We should really be certain (should we?) that we offer original author names at the time publication is created, as this is very important, but being able to link to last version of the person name - if there had been a change in a meantime - and if the system was somehow informed on that change