Arabic Analysis Plugin for Elasticsearch

Arabic Analysis plugin for Elasticsearch. It uses lucene-arabic-analyzer to extract arabic token roots.

Features

• Normalizes input text by removing diacritics and Hamza-like characters
• Extracts word's roots.

Components

• arabic-root analyzer

Example

GET _analyze
{
  "analyzer": "arabic-root",
  "text": "اهْدِنَا الصِّرَاطَ الْمُسْتَقِيمَ"
}

// Result:
['هدن','هدي','صرط','قوم']

Configuration

This plugin is preconfigured with builtin normalization, stop-words and a stemmer which is derived from lucene-arabic-analyzer.

Plugin in action

1. Build the plugin:

mvn clean package

2. Run Elasticsearch and install plugin inside a docker container:

docker compose up

3. Open http://localhost:5601/ and login with elastic/elastic credentials.
4. Go to Dev Tools and examine the plugin:

GET _analyze
{
  "text": "اهْدِنَا الصِّرَاطَ الْمُسْتَقِيمَ",
  "analyzer": "arabic-root"
}