Here is Finnish default config for text field, copied from Dutch and altered to Finnish. Seems to work quite well but Snowball filter may not work perfectly for Finnish stemming.

CommentFileSizeAuthor
#6 2821698-6_finnish_default_config.patch97.04 KBmerilainen
#5 2821698-5_finnish_default_config.patch8.33 KBmerilainen
#3 2821698-3_finnish_default_config.patch8.55 KBmerilainen
#2 2821698_finnish_default_config.patch8.44 KBmerilainen

Comments

mErilainen created an issue. See original summary.

merilainen’s picture

merilainen commented
Status: Active » Needs review
StatusFileSize
new 2821698_finnish_default_config.patch8.44 KB
merilainen’s picture

merilainen commented
StatusFileSize
new 2821698-3_finnish_default_config.patch8.55 KB

Here is an updated patch with Edge NGram filter for better compound word matching.

mkalkbrenner’s picture

mkalkbrenner
Primary language German
Location 🇩🇪
commented
Status: Needs review » Needs work

Thanks for this great start! I wish others would do the same for other languages :-)
Here are some quick notes

  1. +++ b/config/optional/search_api_solr_multilingual.solr_field_type.text_fi_4_5_0.yml
@@ -0,0 +1,493 @@
+          class: solr.MappingCharFilterFactory
+          mapping: accents_fi.txt
...
+    hänessä
...
+    # ä => a
+    "\u00E4" => "a"

    I think there's an error in the char mapping. At least "ä" should be preserved and not be normalized to "a". 

  2. +++ b/config/optional/search_api_solr_multilingual.solr_field_type.text_fi_4_5_0.yml
@@ -0,0 +1,493 @@
+          class: solr.DictionaryCompoundWordTokenFilterFactory
+          dictionary: nouns_fi.txt
...
+  nouns: |

    I don't know if Finish has compound words like German, but the nouns list must not be empty if you keep this filter.

merilainen’s picture

merilainen commented
Status: Needs work » Needs review
StatusFileSize
new 2821698-5_finnish_default_config.patch8.33 KB

I fixed the accents. I know the compound word filter requires a dictionary file, but NL didn't have any either and it won't break Solr. Edge NGram helps a little bit with the compound words, but it won't find them in the middle of the word. Anyway, here is the config without compound word filter. Maybe I will provide a better one if I come up with a better solution.

merilainen’s picture

merilainen commented
StatusFileSize
new 2821698-6_finnish_default_config.patch97.04 KB

Compound filter made a comeback, here with the list of nouns.

mkalkbrenner’s picture

mkalkbrenner
Primary language German
Location 🇩🇪
commented

I applied some changes on commit. Thanks for your contribution!

kekkis’s picture

kekkis
he/him
Primary language Finnish
Location Pirkkala
commented

I wonder, why has this been committed if it's still "Needs review"? Methinks the correct status would be "Fixed".

mkalkbrenner’s picture

mkalkbrenner
Primary language German
Location 🇩🇪
commented
Status: Needs review » Fixed

yep, you're right. thanks

Status: Fixed » Closed (fixed)

Automatically closed - issue fixed for 2 weeks with no activity.