- Article Type: General
- Product: Aleph
- Product Version: 18.01
On April 10, we posted a problem with a long MARC record that had been loaded with a Suppressed status and which we were unable to un-suppress. We followed the recommendation to delete this record and re-load it without the Suppressed status. This worked, but only to a certain extent. Here are our actions and the results.
1. April 4: loaded Aleph #1320664 with suppressed status - functioned as suppressed.
2. April 4: removed suppressed status - continued to function as suppressed.
3. April 10: #1320664 still functions as suppressed; posted support incident #8192-58688
4. April 11: deleted Aleph #1320664 and loaded Aleph #1321752 without suppressed status - functions as un-suppressed. Also noted that some (not all) fields from deleted #1320664 were still keyword searchable.
5. April 18: Suppressed #1321752 - functioned as suppressed very quickly.
6. April 18: Un-suppressed #1321752 - continued to function as suppressed.
7. April 18: Removed all box-level 505 fields from #1321752, resulting in a brief bib record, but record continued to function as suppressed.
8. April 25: #1321752 continues to function as suppressed; words from deleted record #1320664 and fields that were deleted from #1321752 continue to be keyword searchable.
The upshot - we are reluctant to load more of these manuscript records (or other long MARC records) until we have a better understanding of Aleph's indexing behavior. At this point, it seems that all fields are being indexed, regardless of record length, when a record is first loaded into the system, but the indexing that occurs when a record is saved to a server is apparently less comprehensive. Loading these long records as un-suppressed seems to work initially, but it's a short-term solution; if we need to edit these records for whatever reason later will our editing changes take effect? How do we get rid of the ghost keyword index entries from these records?
The limit of index words in each record is 4950 and this can't be changed. So if you define a keyword index on 5XX fields, each field is indexed several times by WRD, WNT, etc., and the record happens to have long 5xx, you'll probably see the problem. One suggestion is to recheck index codes and fields in tab11_word so that each word won't be indexed several times in different index.
Regarding the index of suppressed record 1321752, try to unsuppress it, keep only few fields and save it to server. Then add all fields back. If this doesn't help, you have to wait for the run of p-manage-01. The index of the deleted record will clear up when you rerun p-manage-01.
From Ex Libris Development: In v.22 the limit for word indexing will be increased to 20,000 words (from 4,950). See also, Article 0000040332 ( https://exlibrisgroup--c.na26.visual.force.com/apex/VF_ArticleView?id=kA3320000004LwB&srKp=ka3&srPos=0 ).
- Article last edited: 3/26/2014