TextANN: An Improved Text Classification Model Based on Data Augmentation

This paper explores a new way to improve the effect of text classification. We obtained the session text data which has been desensitized from the online customer service of Alibaba Group and assigned the category label corresponding to each piece of text, indicating whether a piece of text belongs to the session with the system timeout disconnected. The experimental results show that the TextANN provided by us is superior to the traditional text classification model in accuracy, recall and F1-score.