Performance Comparison of Machine Learning Algorithms for Morphological Analysis of Maithili Language

Maithili is an Indo-Aryan language spoken by approximately 36 million speakers in Bihar and Nepal. Presently very less work available for the language. Morphological analyzers are the important component of NLP, which is the motivation towards development of the tool for the computational recognition of language. In this paper we work towards development of morphological analyzer of resource poor language Maithili. We have created dataset of Maithili for morphological analysis, having morpheme and parts of speech classes for the words of the language. We have compared performance parameters like precision, recall, F-measure and accuracy using nine different machine learning algorithms. By comparing all we have found that Instance based learning algorithm is most suitable for designing morphological analyzer for Maithili language.