Analysis of micro-blog comments tendentious vector based on machine learning

At present, the most common communication way in society is constantly developing and expanding. And the number of Internet users doubly increases each year. In this paper, the research object was the tendentious analysis of micro-blog comment information, AdaBoost, Random Subspace, fusion classifier combination method were used to enhance the traditional machine learning method, thereby improving the accuracy of comment analysis. By extracting and fusing several characteristics, the classification effect was enhanced through improved machine learning method, then, the effective fusion of features combined with the naive Bayes, threshold division and other technical methods to determine the spam comments, objective comments. The final experimental results show that using LDA model to extract the theme keyword has certain directional function for the public opinion analysis.