This paper proposes a text data augmentation method that combines TF-IDF and synonym replacement. Firstly, based on the idea that the more frequently a word appears, the more important it is, the frequency of all words appearing in a sentence is counted. However, the same word is likely to...