Method Of Chinese Microblog Sentiment Analysis And Sentiment Element Extraction

Microblog Sentiment Analysis and Sentiment Element Extraction are becoming research hotspots in the field of Chinese natural language processing because of the wide range of applications in public opinion monitor and user experience collection and its short-text and social-network features.The traditional text sentiment analysis methods based on sentiment dictionary or machine learning are lack of suitability to the microblog text features. This article introduces the sentiment analysis dictionaries to research the improved support vector machine classification method based on sentiment analysis dictionaries and the feature combination. In addition, the conditional random field theory is combined with the syntactic dependency parsing to extract the sentiment elements in the microblog more exactly.The sentiment classification method based on sentiment analysis dictionaries and feature combination in this article integrates the advantages of the sentiment classification method based on sentiment dictionaries and the method based on machine learning. The sentiment element extraction method in this article lowers the interference of the short-text and colloquial microblog features in some degree. The main research work is shown as below.(1)First, the research status of different tasks in text sentiment analysis and the text application environment of microblog are introduced. On this basis, the improved Chinese microblog sentiment analysis methods for now are summarized.(2)Second, the sentiment analysis dictionaries are introduced and the polar intensity calculation method for the sentiment words is proposed. The final sentiment value and sentiment polarity of the microblog are acquired on this basis.(3)Then, the conditional random field theory and the syntactic dependency parsing are used to extract the microblog opinion targets. The final sentiment element extraction system can effectively detect the opinion targets and new sentiment words.(4)Finally, the language features of microblog are used in the tradition sentiment analysis algorithm based on support vector machine. The different sentiment feature build modes are used to get more accurate classification performance. The different classification tactics and feature selection algorithms are also used to improve the performance.