Sentiment Classification of Chinese Microblogging Texts with Global RNN

Abstract

Microblogging websites such as twitter and Sina Weibo have attracted many users to share their experiences and express their opinions on a variety of topics. Sentiment classification of microblogging texts is of great significance in analyzing users' opinion on products, persons and hot topics. However, conventional bag-of-words-based sentiment classification methods may meet some problems in processing Chinese microblogging texts because they does not consider semantic meanings of texts. In this paper, we proposed a global RNN-based sentiment method, which use the outputs of all the time-steps as features to extract the global information of texts, for sentiment classification of Chinese microblogging texts and explored different RNN-models. The experiments on two Chinese microblogging datasets show that the proposed method achieves better performance than conventional bag-of-words-based methods.