A structural rule-based stemmer for Persian

Abstract

This paper presents a new stemmer for Persian language. We used a structural approach for stemming which uses the structure of words and morphological rules of the language to recognize the stem of each word. We composed 33 rules to describe a structural rule-based stemmer. The rules are written based on the morphology of Persian language and its word derivation structure. For evaluation, we used our stemmer in an information retrieval system. The results demonstrated that by enhancing the system with this stemmer, the information retrieval system's precision increases, by the factor of 4.78% and the indexing file size decreases by the factor of 6%.