Urdu Stemmer - Rule Based

Urdu Stemmer-Rule Based

Stemming is the process in which inflected words are reduced to find stem or root. There are

various inflected words that can be reduced to stem.

e.g. In English language : 
1) Act can have inflected words like actor, acted, acting etc.
2) Words like fishing, fished and fisher can be reduced to root word fish.

Similarly in Urdu various possibilities have been identified and rules have been developed 
appropriate :

Inflected Word         Root Word 
ںایکڑل                            یکڑل
ںایتسب                            یتسب
ںایڑاگ                            یڑاگ
ںیباتک                            باتک
ےلیم                               لایم

 Approaches

Stemming algorithms are classified under three categories- Rule Based, Statistical and Hybrid.

1) Rule Based approach - This approach applies a set of transformation rules to inflected words 

in order to cut prefixes or suffixes. 

E.g. if the word ends in 'ed', remove the 'ed'. 

2) Statistical approach - The major drawback of Rule Based approach is that it is dependent on 

database. Statistical algorithms overcome this problem by finding distributions of root elements 

in a database. There is no need to maintain the database.        

3) Hybrid approach - It is combination of both Affix removal and Statistical approach. 

Stemming is useful in Natural Language Processing problems like search engine, word 

processing problems and information retrieval. In this stemmer we have applied Rule Based

Approach in which we apply rules on various possibilities of inflected words to remove suffixes 

or prefixes. In Urdu, the only stemmer available to us is Assas-Band developed by NUCES, 

Pakistan which maintains an Affix Exception List and works according to the algorithm to

remove inflections. 

For More details you may read our research paper.

http://aclweb.org/anthology//C/C12/C12-3034.pdf

To Test or Use our Urdu Stemmer please fallow this link.

http://h2p.learnpunjabi.org/ust.aspx