Emotional component analysis and forecast public opinion on micro-blog posts based on maximum entropy model

As the main carrier and platform of spreading network public opinion, micro-blog makes information disseminate more quickly and the influence of public opinion increased. Therefore, accurate analysis and prediction of micro-blog emotion are of great significance for predicting and controlling public opinion. In this paper, we propose the emotional component analysis and public opinion forecast on Chinese micro-blog posts based on maximum entropy model, which uses fine-grained to classify emotion of Chinese micro-blog. Firstly, we preprocess the Chinese micro-blog to filter the noise data. Moreover, the document frequency method and information gain principle are combined to extract features. Secondly, the maximum entropy model is employed to train classifier, and the selective integrated classifiers are used to analyze emotion. On this basis, the principle of the minority subordinate to the majority is used to predict public opinion. In addition, the experimental results have shown the accuracy of the proposed algorithm is 0.88, and the comparison of the four indicators of accuracy, recall, F-Measure and convergence error verify the feasibility and effectiveness of the proposed method.