Skip to Main Content
With the development of World Wide Web technologies, more and more netizens express their opinions on society and politics in net news comments. Sentiment classification is one of the most important sub-problems of opinion mining, which can classify net news comments as positive or negative to help government automatically identify the netizens' viewpoints on news event and make right decision or help enterprises find out weather the customers satisfy the products or not. Most of the researches for sentiment classification only use single classifier, such as kNN, Naive Bayes and Support Vector Machine (SVM). In this paper, we use two multiple classifiers integration algorithms, which are Bagging and Boosting, to conduct the sentiment classification. Different feature selection methods are also investigated. The result of experiment shows that AdaBoost approach, a type of Boosting, usually achieve better performance than Bagging and single classifier and feature selection based on statistic is better than POS-based method for sentiment classification of Chinese net news comments.