[jira] [Commented] (LUCENE-8353) FrenchLightStemmer dont work with ë, ö and ï

["Adrien Grand (JIRA)" <ji...@apache.org>]

    [ https://issues.apache.org/jira/browse/LUCENE-8353?page=com.atlassian.jira.plugin.system.issuetabpanels:comment-tabpanel&focusedCommentId=16509321#comment-16509321 ] 

Adrien Grand commented on LUCENE-8353:
--------------------------------------

This stemmer is borrowed from work from Jacques Savoy (http://members.unine.ch/jacques.savoy/clef/frenchStemmerPlus.txt) so we would diverge from it though I can't think of a reason to not do it, especially given the name of the function.

I wonder how best to handle backward compatibility for existing indices here. Maybe we could fork the stemmer into a new class that has your changes, deprecate the old one, and make FrenchLightStemmerFactory use the new one depending on {{getLuceneMatchVersion}}. Another option could be to add a version parameter to the constructor but it proved problematic in the past (LUCENE-5859). Could you also put comments that make it clear where this stemmer differs from the original one?

> FrenchLightStemmer dont work with ë, ö and ï
> --------------------------------------------
>
>                 Key: LUCENE-8353
>                 URL: https://issues.apache.org/jira/browse/LUCENE-8353
>             Project: Lucene - Core
>          Issue Type: Improvement
>          Components: modules/analysis
>            Reporter: Bruno CAILLAUD
>            Priority: Major
>          Time Spent: 10m
>  Remaining Estimate: 0h
>
> ë, ö and ï are not present in FrenchLightStemmer so if you search per exemple
> Laicité when you index contains Laïcité, that's not match
> I try to fixe this issue in https://github.com/apache/lucene-solr/pull/379



--
This message was sent by Atlassian JIRA
(v7.6.3#76005)

---------------------------------------------------------------------
To unsubscribe, e-mail: dev-unsubscribe@lucene.apache.org
For additional commands, e-mail: dev-help@lucene.apache.org