Malayalam Word Sense Disambiguation using Yamcha

Abstract

Word Sense Disambiguation (WSD) aims to classify each ambiguous word in a particular context with a set of predefined classes. The sense of a specific word in the particular context gives a significant amount of information about the word and its neighbours which can be useful in a language model for different speech and text processing applications. There have been a number of major advances in WSD for many languages, from dictionary-based methods to supervised learning methods and unsupervised learning. This paper describes a hybrid approach using a multi-class SVM and corpus based to Malayalam word sense tagging. This framework makes use of the contextual feature information along with the parts of speech tag feature in order to predict the various WSD classes. For training set, limited number of ambiguous words has been annotated with 16 WSD classes. The experimental results of the 10 fold cross validation shows the appropriateness of the proposed multi-class SVM of Malayalam word sense tagger with one against one approach for both word only and word +POS.