Document-level sentiment analysis aims to predict the overall ratings of texts (e.g., reviews) written by users for items. Most current works model this problem as a supervised learning task, i.e., ...

Most text sentiment classification methods have the shortcomings of insufficient semantic understanding and low classification accuracy. To solve these problems, a sentiment classification model based ...