Research Of Subjective And Objective Classification Method Of Microblog Based On Fusion Features

More and more users prefer to share their views or express their feelings in real-time through microblog, so opinion mining or sentiment analysis of microblog has become a research focus. While the research of subjective and objective classification of microblog is a key work for the research of opinion mining or sentiment analysis, and its main task is to distinguish the microblog contents of express subjective opinion and objective facts, further to tap the valuable information which hidden potential in microblog. In addition, the subjective and objective classification of microblog has great significance for opinion answering system, opinion summary and so on.According to the subjective and objective classification problem of Chinese microblog, this paper studied the basic features combined with different feature selection methods for the effects of the subjective and objective classification result of microblog. While the paper also proposed a subjective and objective classification method of microblog based on fusion features. The main results are as follows:(1) For the grammatical features, this paper proposed a extraction method for words and parts of speech features based on 2-gram. The paper extracted the 2-word and 2-pos features as grammatical features which drawed the 2-gram model for the subjective and objective classification research of microblog.(2) For the semantic features, this paper took full account of the experience of sentiment analysis and its own characteristics of microblog and proposed content features, weighting features and other rich semantic features. The paper also introduced sentiment dictionaries to research the subjective and objective classification.(3) For the problem of feature selection, this paper used two types feature selection methods to compare their classification performance. This paper used different feature selection methods to evaluate the basic features to obtain the optimal features sets, and then combined the classification models to compare the classification results.(4) For the subjective and objective classification problem, this paper proposed a subjective and objective classification method of microblog based on fusion features. The method designed a kind of features fusion algorithm which through combined with the different feature selection methods to obtain effective fusion features, and then combined with machine learning methods to research the classification of subjective and objective.This paper constructs richer subjective and objective classification features, and designs a feature fusion algorithm to explore the effects on subjective and objective classification results after feature selection method combinations. Experimental results show that the proposed feature fusion algorithm can effectively improve the subjective and objective classification results, and the paper builds a relatively common subjective and objective classification model.