Myanmar Dialogue Act Recognition (MDAR)

Abstract

This research aim to make the very first machine learning based Myanmar Dialog Act Recognition (MDAR) for Myanmar Dialogue System. As we know, Dialog Act (DA) recognition is the early level of dialogue understanding which can capture aspects of the user, and they are sentence-level units that represent states of a dialogue, such as greeting, question, inform, and so on. We focus on the current works about DA recognition, especially for Myanmar Dialogue. In this work, we used two machine learning approaches, which are Naïve Bayes classifier and Support Vector Machine (SVM), for dialogue act tagging in the MmTravel (Myanmar Travel) corpus, and the results of two approaches are slightly different but the result of SVM approach attained in the term of average F-measure scores of 0.79; showed that these approach has moderately good accuracy for Myanmar dialogue.