KimbaLtStemmer plugin for ElasticSearch

In order to build KimbaLtStemmer plugin just clone this git repository and initiate maven package build from scratch with this command: mvn clean package.

It's in development stage. (official released version doesn't exist at a time)

Examples:

-----------------------------------------
| Word                 | KimbaLtStemmer |
-----------------------------------------
| taves (singular)     | tav            |
-----------------------------------------
| mūsų (plural)        | mus            |
-----------------------------------------
| namas (singular)     | nam            |
-----------------------------------------
| giedraičiai (plural) | giedraic       |
-----------------------------------------
| geriausias (singular)| geriaus        |
-----------------------------------------
| didysis (singular)   | didys          |
-----------------------------------------

In order to install the latest version of the plugin, simply run:

sudo bin/plugin -url file:elasticsearch-kimba-ltstemmer-0.0.1.zip  -install kimba-ltstemmer

Example usage

Creating index

curl -XPUT http://localhost:9200/test_lt -d '{
  "settings":{
    "analysis":{
      "analyzer":{
        "lt_analyzer":{
          "type":"custom",
          "tokenizer":"standard",
          "filter": ["icu_folding", "stem_lt"]
        }
      },
      "filter": {
        "stem_lt": {
          "type": "KimbaLtStemmer"
        }
      }
    }
  }
}'

Testing analyzer

curl -XGET http://localhost:9200/test_lt/_analyze?analyzer=lt_analyzer&text=Giedraičiai&pretty

And you should get:

tokens: [{
    token: giedraic
    start_offset: 0
    end_offset: 11
    type: <ALPHANUM>
    position: 1
}]

Deleting test index

curl -XDELETE http://localhost:9200/test_lt

Example usage YML configuration

index:
  analysis:
    analyzer:
      lt_analyzer:
        filter: icu_folding, stem_lt
    filter:
      stem_lt:
        type: KimbaLtStemmer

Warning

Input is expected to to be casefolded for Lithuanian, and with diacritics removed. This can be achieved with ICU_FOLDING.