| In recent years, microblog has greatly enriched people’s life. Due to its brief writing, convenient publishing and real-time interacting, microblog becomes very popular. More and more people are actively sharing information with others and expressing their opinions and feelings on microblog. Analyzing emotion hidden in these information can benefit microblog marketing, branding, customer relationship management and monitoring public opinions. Currently, most of the emotional analysis is on English microblog, while Chinese microblog emotional analysis is still at the initial stage.Emotional analysis is to identify the emotional tendencies of the microblog messages, that is to classify users’emotions into positive, negative and neutral. By learning from the traditional text emotional analysis, we analyze the emotional tendencies of microblog based on the characteristics of Chinese microblog. Firstly, summarize and organize the existing resources. Then, use the extended Semantic Orientation Pointwise Mutual Information (SO-PMI) to perform experiments on the Sina microblog and build an emotional dictionary for Chinese microblog. Based on the diversity of expression forms of Chinese microblog, we conduct some preprocessing on the microblog text. We use the emotional words in the microblog text as feature selection method, and process the negative words, adverbs of degree, exclamatory sentence, rhetorical question, and emotional signs in the microblog accordingly. And finally obtain the emotional tendencies by computing the weighted sum of various aspects of microblog messages. A system of emotional tendencies analysis for Chinese microblog is implemented in this paper.The experimental data is selected from Sina microblog corpus in datatang. We manually annotate the microblog messages in domains of science and technology, sports, entertainment. Experimental results show that the method can achieve the accuracy up to74.2%, and the average accuracy is70.5%. The experiment validates the effectiveness of our method, by which we have performed a preliminary exploration of the emotional tendencies analysis of Chinese microblog in this paper. |
