Same kind of bug as 9352, for 5xx fields (titles) Zebra indexes all subfields of UNIMARC 5XX fields in title index. It can cause Koha to give too much results. M. Saby
Created attachment 16176 [details] [review] [PATCH] Bug 9821: Make indexing of UNIMARC 5XX and 205 more specific 511 field ("false title") is not indexed. 519 field is indexed but does not belong to unimarc standard. Indexing the whole 500/501 make search noisy because of $k (date) and $m (language). $b is useless. Indexing some subfields of 520/530/532/560 could sometimes be harmfull too. In 500/501, $j,$x,$y,$z are subdivisions used of the record is not about a title, but about a subject linked to a title. For example : 500$aBible$xTranslations So they should be indexed in subjects indexes rather than title index. Regarding 205 (edition statement), it is currently indexed in title index. But some subfields of 205 ($f and $g) contain information about the authors of that edition, so only $a,$b,$d should be indexed in title index. An other patch is made for indexing $f $g in author index. This patch does the following changes : 1) Indexing only 205$a,$b,$d in title index 2) Commenting all the 5XX in record.abs 3) Indexing 511 4) Keeping 519 for the moment, but commenting 5) For 500 and 501, indexing subject subfield in subjects index : 500/501$j Subject,Subject:p 500/501$x Subject,Subject:p 500/501$y Subject-name-geographical:w,Subject-name-geographical:p,Subject,Subject:p 500/501$z Subject-chronological:w,Subject-chronological:p,Subject,Subject:p 6) Unindexing the following subfields : 500 and 501 (uniform title) not indexing $b (type of ressource), $k (date), $m (language), $2 (system code),$3 (authority record number) So that 500 $aBible $iNew Testament $mEnglish $qRevised Standard Version $k1972 should be indexed as 500 $aBible $iNew Testament $qRevised Standard Version 520 : preceding title (used for serials) non indexing $j and $x $j is intended for displaying notes. Ex of content : "Issue nos. 1 (summer 1974)-5 (autumn 1975)" $x contains ISSN of former title. Maybe it could be indexed in issn index, but it should not be in Title 530 : key title (used for serials) non indexing $j $j contains volume and date of key title. Could be used if the record contains several 530, to precise the volumes or dates associated with each key title. Nobody will use it in ordinary search. 532 : Developped Title non indexing $z (language of title) 560 : Artificial Title not indexing $5 (code for institution to which the field applies) To test : In a GRS1 environment 1) Create a record with 511$a FalseTitle 205$a Edition $f bz9821 $g bz9821 500$a Bible $b bz9821 $k bz9821 $m bz9821 $j bz9821subject $x bz9821subject $y bz9821subject $z bz9821subject $2 bz9821 $3 bz9821 501$a Bible $b bz9821 $k bz9821 $m bz9821 $j bz9821subject $x bz9821subject $y bz9821subject $z bz9821subject $2 bz9821 $3 bz9821 520$a PTitle $x bz9821 $j bz9821 530$a KTitle $v bz9821 532$a DTitle $z bz9821 560$a DTitle $5 bz9821 2) Apply 3) Rebuild zebra 4) Search in title index : FalseTitle => you should find it 5) Search in title index : bz9821* => you should have no result 6) Search in subject index : bz9821subject => you should have 1 result Do the same in a DOM environment M. Saby
See also 9352 for 7XX fields and 8252 for 1XX fields. Still some corrections to do for 9XX fields and 6XX fields. M. Saby
Please don't test that for the moment. I fear I made some mistakes :( I will rewrite it when other bugs about UNIMARC are passed. M. Saby
I think we can safely kill this one off.. it's not been looked at for 5 years!