Machine learning-based judgment of text emotional relevance. Train machine learning models to determine whether news text contains political propaganda. The main work includes: • Participate in manual annotation of training set data • Responsible for the preprocessing of some text data: data cleaning (removing punctuation marks, stop words, etc.), word segmentation processing, and stemming • Participate in the construction of bag-of-words model and vectorize text data • Use multiple machine learning algorithm models such as SVM, LR, RF, NB, KNN, etc. to train the training set and test the prediction performance of each model