Sentiment Analysis of Emergencies Based on Microblogging

摘要

With wide use of microblogging, the sentiment analysis of emergencies based on microblogging is helpful to analyze the trend of public opinion, and is beneficial to monitor and guide the public opinion correctly for government. Firstly, word2vec is used to transform microblogging text into the feature vector with high-dimensional space. Then, a classification algorithm based on random forest optimized by genetic algorithm is proposed. Finally, an experiment is performed. The result shows that the accuracy of microblogging sentiment classification based on proposed classification algorithm is improved greatly compared with classical single classifiers.