The paper describes a new stemmer algorithm to find the roots and patterns for Arabic words based on excessive letter locations. The algorithm locates the trilateral root, quadri-literal root as well as the pentaliteral root. The algorithm is written with the goal of supporting natural language processing programs such as parsers and information retrieval systems. The algorithm has been tested on thousands of Arabic words. Results reveals an accuracy reached to 95%.
展开▼
