Hi, I have been doing some work with the (Zebra) indexing definitions for MARC21 authority records. One of the current problems is that it is impossible to effectively search on the complete heading if subdivisions are present. For example, a topical heading such as
150 $a Medicine $x Ability testing is currently indexed as two separate phrases, "Medicine" and "Ability testing". A search for "medicine ability testing" doesn't work, and searching on the two phrases separately (he,ext=medicine and he,ext="ability testing") would also pick up a "ability testing -- medicine" heading if such existed in the authority file. I have found no way to fix this using Zebra's GRS-1 filter and its elm and melm directives in records.abs. Consequently, I have been working with Zebra's DOM XML filter, which generates index entries by mapping a MARC XML record to index terms via XSLT, and have been able to get much better results. Two patches containing the XSLT can be found at http://manage-gmc.dev.kohalibrary.com/patches/ Since manually maintaining the whole indexing XSL would be rather unwieldy, I've created a concise XML representation of the indexing definitions, authority-koha-indexdefs.xml. This in turn is mapped via the XSL transform koha-indexdefs-to-zebra.xsl to produce the XSL to be used by Zebra, authority-zebra-indexdefs.xsl. I have not yet made a formal schema for the Koha indexing definitions XML, but once its syntax has been extended to cover UNIMARC and bibliographic records, this could be an approach to take to have all Zebra indexing use the DOM XML module. In addition, it opens the possibility of creating code tthat allows users to work with the Zebra indexing definitions from the staff interface -- the definitions could be stored in the database, then serialized to XML and transformed to the Zebra indexing XSLT. Comments requested. Regards, Galen -- Galen Charlton Koha Application Developer LibLime [EMAIL PROTECTED] p: 1-888-564-2457 x709 _______________________________________________ Koha-devel mailing list [email protected] http://lists.nongnu.org/mailman/listinfo/koha-devel
