Sentiment Tendency Analysis Of The Chinese Microblog Comment

With the rapid development and the popularization of Internet, more and more peopleuse Microblog. People become interested in expressing their own opinions and commentsabout something on Microblog, so Microblog becomes the bridge for communicating withpeople, and it becomes one of the most important social platforms. However, the research ofthe sentiment analysis on Chinese Microblog is relatively less than the social platforms suchas Twitter and Facebook. The review data commented by users showed the attitude about theproduct or the information resource to a certain extent, and showed the sentiment tendencystrongly at the same time, so the sentiment tendency analysis has gradually become the hotspot of the research of the natural language processing. According to the research of the textgranularity, sentiment analysis is usually divided into word and phrase sentiment analysis,sentence sentiment analysis, text sentiment analysis. The main research work of this paper isas follows:(1) The current research status,research background and research significance of thesentiment analysis are introduced.(2) Detail introduce of the related basic knowledge in sentiment tendency analysis. In theview of the existing Chinese vocabulary words polar sememe cannot consider the meaningcompletely, and lead to inaccurate polarity analysis, aiming at these shortage, a higherpropensity of a lexical semantic precision calculations is proposed. We should firstly find a setof mixed neutral library, and extract the testing word and the polarity meaning, suffix, othermeanings of the standard words, then calculate their similarity by giving appropriate weightvalue, lastly we can obtain the emotion tendency of the test words.(3) According to the research on the existing sentiment classifications of the text, wemake a comparative analysis on their respective characteristics and problems of thesemethods. In the view of the existing problems, and compared with the traditional Bayesalgorithm and SVM algorithm, this paper proposes an algorithm of the paragraph sentimenttendency analysis based on the ANN. Through the experimental analysis we can find that thealgorithm of this paper has certain increase in the accuracy compared with the Bayesalgorithm, but still worse than the SVM algorithm, this method needs to be further research.