A new stemmer for Nepali language

Nepali is an inflectionally rich language. The unavailability of a stemmer, that effectively eliminates inflections, makes the task of searching, text mining, information retrieval and natural language processing in Nepali challenging. Stemming helps in mapping all variants of inflected words to their respective stems. Stemmers are widely used in Text Mining, Information Retrieval and Natural Language Processing systems to improve the performance of such systems. This paper presents a new rule-based stemmer for Nepali language. We have composed 128 suffix rules, which are executed in step-by-step and iterative manner to eliminate inflections. The proposed stemmer was tested on 5000 Nepali words and the overall accuracy was around 88.78%.