Negative review detection model based on LightGBM

Abstract

With the development of the Internet, online comments can be seen everywhere on major social platforms, and their content also involves all aspects of people’s life, such as clothing, food, housing, and transportation. However, it contains a large number of illegal negative comments released by Internet navy, these negative comments are often issued by Internet navy hired by Internet public relations companies, disrupting the order of the Internet, creating Internet panic, and seriously affecting public opinion. In this paper, the LightGBM algorithm is used, and the bag of words and the IF-IDF model are used for feature extraction to construct an illegal negative comment recognition model. By training with the IDMB training set, the results show that our model is more than 90% accurate. And compared to the popular classification models, our model has higher performance.