With the development of Web 2.0, more and more users are happy to share their opinions and experiences on the internet. Subsequently, it is increasingly difficult for people to collect and process the huge information from the network. Therefore, text sentiment classification based on the computer is proposed to tackle this problem. And one of the most important research directions is to enhance the classification accuracy for text sentiment classification. In addition, ensemble learning is an effective approach to enhance the classification accuracy and has shown better performance than base classifiers in many fields. Based on these considerations, text sentiment classification based on ensemble learning is proposed to enhance the performance of classifiers. Experimental results reveal that three ensemble methods, i.e., Bagging, Boosting and Random Subspace, enhance the classification accuracy of different base classifiers. Compared with Bagging and Boosting, Random Subspace gets more significant improvement of the classification accuracy. All these results demonstrate the effectiveness and feasibility of application of ensemble learning in text sentiment classification.