I have a friend that is new to Koha and I've been able to help up until now - so I thought I'd turn to you all for help. She's having issues with rebuilding authorities. She's on version 3.0.5. She follows these steps: 1) Export records from our ILS and use MarcEdit to get them like Koha wants them 2) Use bulkmarcimport.pl and bulkauthimport.pl to load records 3) Run link_bibs_to_authorities.pl 4) Run rebuild_zebra.pl Two problems with authority linking: When I'm done with these steps, I can see all loaded bib, item, and authority records in my system. And, some (but not all) of the authority records are linked to their related bib records (problem 1). But, when I try to do an author search on a name in an existing authority record but not in one of its linked bib records, no results are returned (problem 2). And a tertiary problem: I've discovered that some of my records contain errors that cause the bulkimport scripts to fail. I think, but am not sure, that this is related to non-Roman characters in the records. Two questions related to this problem: 1) Is there are trick for identifying these records before I actually try to load them? 2) I made the dumb mistake of double loading about 8000 authority records because I forgot to rebuild_zebra after my authority load script failed. Is there a good way to do one of the following: a) Batch delete all authority records with no bibs attached. b) Batch delete all "duplicate" authority records. c) Wipe out the entire authority table and start from scratch.
1) Export records from our ILS and use MarcEdit to get them like Koha wants them 2) Use bulkmarcimport.pl and bulkauthimport.pl to load records 3) Run link_bibs_to_authorities.pl 4) Run rebuild_zebra.pl
Two problems with authority linking: When I'm done with these steps, I can see all loaded bib, item, and authority records in my system. And, some (but not all) of the authority records are linked to their related bib records (problem 1).
bulkmarcimport.pl is deprecated. You're supposed to use bulkauthimport.pl. Check for your Koha version. A better approach can be to do biblio record to authorities links before importing into Koha. This way you don't have to use link_bibs_to_authorities.pl which isn't an accurate method to do linking. This script can't deal with homonymy. You can even keep authorities ID from the legacy ILS you're coming from. Between 2 and 3, you should have run rebuild_zebra -r -a to reindex authorities. link_bibs_to_authorities.pl try to match authorities controlled biblio fields by searching into Zebra authorities indexes.
But, when I try to do an author search on a name in an existing authority record but not in one of its linked bib records, no results are returned (problem 2).
???
And a tertiary problem: I've discovered that some of my records contain errors that cause the bulkimport scripts to fail. I think, but am not sure, that this is related to non-Roman characters in the records. Two questions related to this problem: 1) Is there are trick for identifying these records before I actually try to load them?
Try to use yaz-marcdump command.
2) I made the dumb mistake of double loading about 8000 authority records because I forgot to rebuild_zebra after my authority load script failed. Is there a good way to do one of the following: a) Batch delete all authority records with no bibs attached. b) Batch delete all "duplicate" authority records. c) Wipe out the entire authority table and start from scratch.
Duplicates authorities is the reason why link_bibs_to_authorities.pl doesn't work properly. It requires authority exact matching. Delete all your authorities and re-import them. For c), in MySQL: TRUNCATE auth_header; -- Frédéric
And a tertiary problem: I've discovered that some of my records contain errors that cause the bulkimport scripts to fail. I think, but am not sure, that this is related to non-Roman characters in the records. Two questions related to this problem: 1) Is there are trick for identifying these records before I actually try to load them?
Try to use yaz-marcdump command.
2) I made the dumb mistake of double loading about 8000 authority records because I forgot to rebuild_zebra after my authority load script failed. Is there a good way to do one of the following: a) Batch delete all authority records with no bibs attached.
There is a way to do this :
mmmm... latest bulkmarcimport should be more failsafe. try remove_unused_authorities.pl in misc/migration_tools It is on the official/master branch It safely removes unused authorities. my 2 cents. -- Henri-Damien LAURENT
participants (3)
-
Frederic Demians -
LAURENT Henri-Damien -
Nicole Engard