Research Of Sentiment Analysis For The Chinese Micro Blog Based On Emotional Dictionary And Sentence Classification

With the development of social network, the way that people get information has changed greatly, more and more people get used to the Internet to get the news, the hot spot information and so on. Among them, Micro Blog as a kind of widely used in network media receives the favor of majority of Internet users by its abundant communication content and means of communication in anytime, anywhere. Simultaneously, large amount of users make their data also showed an exponential growth, which contains a large amount of emotional information. To analyze the sentiment orientation to these data, it is not only can grasping users’ attitudes to major events and business products, but also can provides strong foundation for market research products, network public sentiment warning and government decision-making, which really has enormous social and commercial value.Most studies of existing Micro Blog sentiment analysis are targeted at the English Micro Blog, while the studies of sentiment analysis for Chinese Micro Blog are still in its infancy. This paper mainly conducted the following works for Chinese Micro Blog features:1. Overview of sentiment analysis for Chinese Micro Blog. It introduces the key features of Chinese Micro Blog, analyzes and summarizes the previous theoretical analysis methods of sentiment analysis for Micro Blog. Moreover, comparing the method of sentiment analysis based on the emotional dictionary and method of sentiment analysis based on the machine learning, it proposes a new method of sentiment analysis for Chinese Micro Blog based on emotional dictionary and machine learning.2. Construction of emotional dictionary for the Chinese Micro Blog. It extends the network for emotional lexicon to build emotional dictionary for Chinese Micro Blog based on the existing emotional resources, and describes the method of Construction in detail, such as the basic emotional dictionary, the network emotional dictionary and the expression symbol database. On this basis, it uses the SO-PMI algorithm smoothed by Laplace to expand the Micro Blog emotional dictionary.3. Research of sentiment analysis for the Chinese Micro Blog based on sentence classification. It analyzes the effect of different sentences on sentence sentiment orientation in depth and classifies the sentences by the improved C4.5 decision tree algorithm. Furthermore, analyzing and processing for the different sentence, it calculates the values of sentence sentiment, and takes into account the impact of punctuation, related words, and sentence position for emotional intensity to adjust the emotional weight of the sentences and analyze the sentiment orientation for Chinese Micro Blog.Based on the research above, it uses the Micro Blog standard data sets as test corpus from Task Four of COAE2014, and designs of experiments to verify the feasibility of the proposed method. The results show that this method can be able to analyze the sentiment orientation for Chinese Micro Blog effectively and has a high accuracy rate.