Publication

Abstract : The role of morphological analyzer is very significant in the field of natural language processing (NLP) applications like machine translation (MT), information extraction (IE), information retrieval (IR), spell checker, lexicography etc. So from a serious computational perspective the creation and availability of a morphological analyzer for a language is important. The morphological analyzer maps an inflected word into its stem, parts of speech and feature equations corresponding to inflectional information. The morphological structure of an agglutinative language is unique and capturing its complexity in a machine analyzable and generatable format is a challenging job. This paper presents a paradigm based morphological analyzer, for the complex agglutinative Kannada language using the machine learning approach. The proposed morphological analyzer is designed using sequence labeling approach and training, testing and evaluations are done by support vector method (SVM) algorithms. The system captures the various non-linear relationships and morphological features of Kannada language in a better and simpler way. We also compared the efficiency of our system with the existing morphological analyzers which are publically available in the internet. From the experiment we found that the performance of our system significantly outperforms the existing morphological analyzer and achieves a very competitive accuracy of 96.25% for Kannada verbs.