Combination Model for Sentiment Classification Based on Multi-feature Fusion

Sentiment classification is a way to analyze the subjective information in the text and then mine the opinion. This paper focuses on the word level sentiment classification. A combination model for word level sentiment classification based on multi-feature fusion is proposed in this paper. Firstly, different combinations models of various features are gotten and the Naive Bayes classifier is trained by each of them. Secondly, the results of Chinese sentiment classification for each model are compared and the effective factors to classification are analyzed. Finally, the best combination model can be obtained from the before. This paper aims to explore the best feature combination model to improve classification effectiveness. The experiments show that the classification precision rate and recall rate of proposed model is higher and more optimal and significantly than single feature selection model.