Intelligent environment-friendly propaganda and education method combining text topic analysis, sentiment analysis and GSVM

The invention discloses an intelligent environment-friendly propaganda and education method combining text topic analysis, sentiment analysis and GSVM, and relates to the technical field of Internet, and the method comprises the following steps: for given propaganda tweet data, dividing original data into dichotomous label samples according to a threshold value of 50% quantile, taking the sample set and a plurality of corresponding extracted feature variables (a title text subject, a title emotion, a special symbol, release time, whether a title is a header or not and the like) as main data input and basis to enter a next link, and obtaining the sample set based on the previous step. According to the method, prediction classification can be carried out on data, self-adaptive parameter optimization breaks through the prediction accuracy of existing research, the prediction accuracy is close to 90%, more targeted decision guidance is provided for propagandists, efficient, accurate and practical tweets are provi