Bug 23918 - Search results incorrect for multiple vowel words
Summary: Search results incorrect for multiple vowel words
Status: RESOLVED WONTFIX
Alias: None
Product: Koha
Classification: Unclassified
Component: Searching (show other bugs)
Version: Main
Hardware: All All
: P5 - low minor (vote)
Assignee: Bugs List
QA Contact: Testopia
URL:
Keywords:
Depends on:
Blocks:
 
Reported: 2019-10-29 15:55 UTC by Bernard
Modified: 2023-12-09 12:29 UTC (History)
1 user (show)

See Also:
Change sponsored?: ---
Patch complexity: ---
Documentation contact:
Documentation submission:
Text to go in the release notes:
Version(s) released in:


Attachments

Note You need to log in before you can comment on or make changes to this bug.
Description Bernard 2019-10-29 15:55:05 UTC
On a debian Koha package install without any zebra modifications, we observe searching for words with double vowels is returning odd results.

eg on https://lst.koha-ptfs.eu/cgi-bin/koha/opac-main.pl on Koha 18.1110000

food = 24 results
foood = 22 results

Foood (with 3 consecutive letter o's) should not return results. Also observed on other package install examples.
Comment 1 Katrin Fischer 2019-10-29 21:27:35 UTC
It could be that oo is reduced to o in the Zebra mapping (I had something similar happen with i ii iii, but haven't checked for ooo), so you actually end up with food. Depending on your settings (fuzzy search and similar) this might also be expected, as those are thought to catch simple spelling errors.
Comment 2 Bernard 2019-10-30 12:37:25 UTC
I don't think it is fuzzy search settings. QueryFuzzy is set to "Don't try". Which file in the zebra mapping could it be so I can check?
Comment 3 Katrin Fischer 2019-10-30 19:48:37 UTC
I think it's etc/zebradb/etc/word-phrase-utf.chr
Comment 4 Bernard 2019-11-04 12:51:39 UTC
Thanks. I can't see anything in that file that would map 1 lowercase letter o to 2 (or 2 to 3). Our file is like this: https://github.com/Koha-Community/Koha/blob/18.11.x/etc/zebradb/etc/word-phrase-utf.chr Any idea which line might be causing this to happen?
Comment 5 Katrin Fischer 2019-11-04 22:02:16 UTC
(In reply to Bernard from comment #4)
> Thanks. I can't see anything in that file that would map 1 lowercase letter
> o to 2 (or 2 to 3). Our file is like this:
> https://github.com/Koha-Community/Koha/blob/18.11.x/etc/zebradb/etc/word-
> phrase-utf.chr Any idea which line might be causing this to happen?

My theory is, that the combination of those is doing it:

equivalent ō(oo)
map ō			o

I could be wrong :)
Comment 6 Bernard 2019-11-12 14:14:22 UTC
You were right! Commenting out 
equivalent ō(oo)
did the trick for that example.

Unfortunately, there may be similar problems for other vowel combinations. eg
aa ("aachen" brings up "aachen" and "achen") 


I am assuming 
equivalent ā(aa)

might be the one to comment out for that, but of course it means those diacritics are not properly mapped then if they should ever genuinely come up.
Comment 7 Katrin Fischer 2019-11-13 14:50:10 UTC
(In reply to Bernard from comment #6)
> You were right! Commenting out 
> equivalent ō(oo)
> did the trick for that example.
> 
> Unfortunately, there may be similar problems for other vowel combinations. eg
> aa ("aachen" brings up "aachen" and "achen") 
> 
> 
> I am assuming 
> equivalent ā(aa)
> 
> might be the one to comment out for that, but of course it means those
> diacritics are not properly mapped then if they should ever genuinely come
> up.

A library reported this to us and we decided together that we won't change it... too worried about side effects for other searches, especially if you have records in multiple languages where these might appear...
Comment 8 Katrin Fischer 2020-03-28 15:09:09 UTC
Should we mark this WONTFIX for now?