摘要
With the development of Internet, people are more likely to post and propagate opinions online. Sentiment analysis is then becoming an important challenge to under- stand the polarity beneath these comments. Currently a lot of approaches from natural language processing＇s perspec- tive have been employed to conduct this task. The widely used ones include bag-of-words and semantic oriented analy- sis methods. In this research, we further investigate the struc- tural information among words, phrases and sentences within the comments to conduct the sentiment analysis. The idea is inspired by the fact that the structural information is play- ing important role in identifying the overall statement＇s po- larity. As a result a novel sentiment analysis model is pro- posed based on recurrent neural network, which takes the par- tial document as input and then the next parts to predict the sentiment label distribution rather than the next word. The proposed method learns words representation simultaneously the sentiment distribution. Experimental studies have been conducted on commonly used datasets and the results have shown its promising potential.
With the development of Internet, people are more likely to post and propagate opinions online. Sentiment analysis is then becoming an important challenge to under- stand the polarity beneath these comments. Currently a lot of approaches from natural language processing＇s perspec- tive have been employed to conduct this task. The widely used ones include bag-of-words and semantic oriented analy- sis methods. In this research, we further investigate the struc- tural information among words, phrases and sentences within the comments to conduct the sentiment analysis. The idea is inspired by the fact that the structural information is play- ing important role in identifying the overall statement＇s po- larity. As a result a novel sentiment analysis model is pro- posed based on recurrent neural network, which takes the par- tial document as input and then the next parts to predict the sentiment label distribution rather than the next word. The proposed method learns words representation simultaneously the sentiment distribution. Experimental studies have been conducted on commonly used datasets and the results have shown its promising potential.
基金
This work was partially supported by the Na- tional High Technology Research and Development Program of China （2011AA010502）, the National Natural Science Foundation of China （Grant No. 61103095）, and the Fundamental Research Funds for the Central Uni- versifies. We are grateful to Shenzhen Key Laboratory of Data Vitalization （Smart City） for supporting this research.
