Myanmar news sentiment analyzer using support vector machine algorithm

Abstract

Sentiment analysis is one of the natural language processing fields that combined computational linguistics and information retrieval. It is one of text classification techniques that extract opinion expressed in a positive and negative. The news, blogs and reviews obtained from social network are become important resources for sentiment analysis. This paper implements sentiment analyzer for Myanmar news. This system create sentiment annotated corpus for Myanmar news that are collected from Myanmar news web sites and ALT Treebank. Feature extraction and transformation are also needed to extract feature and transform to feature vector to get high performance. N-gram and TFIDF weighting methods are used in this system. Machine learning is combination of the techniques and basis from both statistics and computer science. This system is implemented by using machine learning algorithm Support Vector Machine compared with Logistics Regression and Naïve Bayes algorithms. We showed the comparison results of those algorithms and also showed that SVM is the more powerful than other two algorithms. User can easily know how importance about news by using this sentiment analyzer system.