Chinese Microblog Public Sentiment Supervisory System Based On Text Sentiment Analysis

With the rapid development of Internet technology and mobile technology, as an important part of the Internet media, micro-blog has been risen up rapidly. Micro-blog has become one of the most popular social networking platform. More and more Internet users use micro-blog to express their views and attitudes. Especially, the mobile phone terminal makes sharing and dissemination of information through micro-blog become more convenient. In this process, different user groups make the micro-blog information diverse.So a large number of individual viewpoints filled the entire Internet. For the government,enterprises, network security and other organizations, these motional information is very important. So the detection and warning of micro-blog social media platform becomes more and more important. Text tendency classification is a method to discovery and analysis the views and position on a certain topic, so we can know the emotional tendency is positive or negative. This technology can be widely used in topic analysis, public opinion analysis,product reviews, market research, public opinion warning, etc.At present, just a few systems use text tendency analysis in micro-blog and micro-blog sentiment analysis method also exists a low accuracy rate problem without comprehensive consideration. What’s more, lack of considering in the aspect of dependency relationship. In terms of the above characteristics, this paper, from the characteristics of micro-blog, which is based on existing research results, improved the existing analysis methods and come up with an effective micro-blog sentiment analysis method. At last, this paper designed and implemented micro-blog monitoring system. The main research contents and innovations in this paper are as follows:1. Researching existing text tendency analysis method, mainly from text tendency analysis method based on semantic. On the one hand, the paper researched and analyzed the pattern which is based on word and phrase. On the other hand, talked about the text tendency analysis method based on semantic pattern base. And also compared and summarized the two methods.2. This paper constructs a set of text emotional dictionary which is required in text tendency analysis.Including domain emotional dictionary, basic emotion dictionary,dictionary of adverb, expression and symbol dictionary, and network words dictionary.Besides, the paper proposed a kind of domain emotional dictionary construction method, so that it can automatically extend emotion lexicon, reduce manual intervention. This paperalso combined HowNet with Chinese appraise dictionary, and proposed a kind of basic dictionary construction method.3. Combining the methods which is mentioned in the first point, this paper proposed text tendency analysis method based on emotional dictionary and phrase patterns. Firstly,analyzed the phrase pattern library. Summed up phrases expressions as seven kinds of phrase patterns. Secondly, analyzed the dependencies between words. Finally, combining emotional dictionary and phrase patterns to calculate emotional tendency of sentences.This paper combines web crawler technology, page analysis technology, text preprocessing technology, and so on. Proposed and designed a set of micro-blog public opinion monitoring system. Including micro-blog acquisition module, analysis module,service module, etc. In the end, the results show that the system meets the expected effect by testing the system.