Assem's Arabic Stemmer

This is an algorithm for Arabic stemming written on Snowball framework language. If offers light stemming and text normalization. voc

Requirements:

• Snowball framework

• Snowball-data

• Golden-Arabic-Corpus

• NLTK

• mkdocs

• You can download it automatically using:

    $ make download

• Install python requirements

    $ sudo pip install -r requirements.txt

or manually by:

• extracting snowball into the root folder {Root}/snowball
• extracting snowball-data/arabic/voc.txt.gz into {Root}/test_data/voc.txt

Build:

• light stemming

      $ make build

• root-based stemming

      $ make build_root_based_stemmer

Run:

• Light Stemmer

  	 $ make run
  	  الطالب
  	  طالب

• Root-Based Stemmer

      $ make run_root
      الطالب
      طلب

Test:

We configured tests to run against snowball-data arabic sample.

• time:

      $ make time

• grouping effect:

      $ make grouping

• all:

      $ make test

• Test SAS with golden arabic corpus:

      $ make test_arabicstemmer

• Test ISRI Stemmer with golden arabic corpus:

     $ make test_isri

Distributions:

• dist light stemmer to available languages:

    $ make dist

• dist root-based stemmer to available languages:

    $ make dist_rooter

Results:

Snowball Arabic (Stemmer & rooter) Results

Word	Stem	root
طفل	طفل	طفل
اطفال	اطفال	طفل
الاطفال	اطفال	طفل
اطفالكم	اطفال	طفل
فأطفالكم	اطفال	طفل
اطفالهم	اطفال	طفل
والاطفال	اطفال	طفل
فاطفالهم	اطفال	طفل
وطفل	طفل	طفل
الطفولة	طفول	طفل
والطفلتين	طفل	طفل
طفلتان	طفل	طفل