Stemming in Tamil for Affix Stripping

Stemming is the one of the most important step in many of the Natural Language processing tasks. Stemming reduces inflected words to a common stem/root word. Stemming process mainly carried out in English language because Tamil language is more complex in structure and more over it consists of critical grammatical rules. Tamil is a Dravidian language, mainly spoken by Tamil. Tamil words have more derivational forms than any other languages. The words get inflected to different forms based on number, person, gender and tense. Taking this complexity into account, the present attempt is to propose the effective stemmer for the Tamil language. The stemmer is used to find out the stem/root word for the Tamil language input. So the stemmer is built based on Affix stripping algorithm for the Tamil input. Then the stemmer reduces over-stemming and under-stemming errors. Affix stripping algorithm removes the prefix and/or suffix from the given Tamil input based on some rules. The input is taken as the Tamil text. Then the raw input is going for the Normalization, Feature extraction and Tokenization process. It removes the special characters, Stop words from the input and divides the input into tokens respectively. Then it fed into the stemmer for the stemming process. It finally displays the collection of stemmed words as an output and the performance, efficiency of the stemmer is evaluated.