Sentiment Analysis(SA) for Kannada documents has been explored recently. In the recent study [8], the sentiment analysis for Kannada text is explored using Naive Bayes classifier. The objective of this work is to improve the performance of the previous study on the sentiment analyzer for Kannada language explored in the paper [8]. In this work, we propose the ensemble of classifier with random forest technique to identify the polarity of the sentiment and test the performance of the same. Also in this work, some of the limitations of [8] such as handling multi class labels, identification of sentiment polarity of comparative and conditional statements have been addressed. The over all accuracy is improved from 65% to 72 %, indicating our approach based on Random Forest technique is more efficient for SA for Kannada.
展开▼
