ParsiAnalyzer is an analysis plugin for Elasticsearch. Analysis is a process that consists of the following steps:

• Tokenizing a block of text into individual terms
• Normalizing these terms into a standard form

An analyzer is really just a wrapper that combines Character filters, Tokenizer, and Token filters. Elasticsearch provides many Built-in Analyzers but there's still room for improvement especially for Persian language. This plugin provides tools for tokenizing, normalizing and stemming Persian text.

• Tokenize Persian text

  • Convert whitespaces to zero width nonjoiner (نیم‌فاصله) whenever it is necessary. for example,می رود to می‌رود.
  • Convert Persian punctuations to their English equivalent. for example,۳/۱۴ to ۳.۱۴
  • Tokenize Persian text by whitespaces and punctuations.

• Normalize Persian tokens into a single canonical form

  • Transform all forms of Yeh, Kaf, Heh, and Hamza to a unique form. for example,براي to برای.
  • Convert all Persian and Arabic numbers to their English equivalent. for example,۱۴۳ to 143.
  • Remove diacritic (اِعراب) from words. for example, اَرّه to اره.
  • Remove Kashida form words. for example, بادبــــــادک to بادبادک.

• Remove common Persian stop words

  • Persian stop words like از, به and etc will be removed.

• Stem Persian words

  • Remove common Persian suffixes. for example, ها or ان.

To install the plugin for Elasticsearch 7.13.1, run this command:

bin\elasticsearch-plugin install https://www.dropbox.com/s/cr61dmnx95taivi/ParsiAnalyzer-7.13.1.zip?dl=1

If you want to build ParsiAnalyzer for any specific version of Elasticsearch, follow these steps:

1. Make sure you've installed JDK and Maven on your computer
2. Clone project
3. Open pom.xml
4. Under dependencies tag, change Elasticsearch version to your desired version
5. Open plugin-descriptor.properties
6. Change elasticsearch.version to your desired version
7. Run this maven command: mvn clean package
8. In the target/releases folder, you’ll now find a zip file. install the plugin using this command: bin/elasticsearch-plugin install file:///path/to/ParsiAnalyzer.zip

To see how this plugin works, you can use Elasticsearch's analyze API:

POST _analyze
{
  "analyzer" : "parsi",
  "text" : "روباه قهوه‌اي چابك از روی سگ تنبل می پرد"
}

If you find stemming a little annoying, you can always use the standard variation of ParsiAnalyzer:

POST _analyze
{
  "analyzer" : "parsi_standard",
  "text" : "روباه قهوه‌اي چابك از روی سگ تنبل می پرد"
}

ParsiAnalyzer can be specified directly in the field mapping as follows:

PUT /my_index
{
  "mappings": {
    "blog": {
      "properties": {
        "title": {
          "type":     "text",
          "analyzer": "parsi" 
        }
      }
    }
  }
}

Email: n.esmaielyfard [at] gmail.com