Enhancing a Greek Language Stemmer - Efficiency and Accuracy Improvements

Abstract

Stemming algorithms are used in the field of Information Retrieval in order to improve precision and recall. Although for Greek there are three stemmers published, only one of them is freely available. In this thesis, we use stemmer performance metrics for evaluating the existing algorithm and we improved its accuracy and completeness. These improvements were achieved by providing an alternative implementation in PHP which offers more syntactical rules and exceptions. Finally, the two algorithms are tested and their statistics metrics are compared.

Key words and terms: stemming algorithm, Greek, algorithmic efficiency, PHP