The project selected topic from the internationalization of China computer society organization intelligent algorithm, innovative applications and big data system large challenge tournament (https://www.datafountain.cn/competitions/350). This task builds a text emotion analysis model based on short Internet texts and their emotion polarity labels (0,1,2) to realize the emotion polarity judgment of new input texts. The data set was provided by the International Intelligent Algorithm, Innovative Application and Big Data System large-scale Challenge Competition organized by China Computer Society. Using Internet online data, the titles and text of more than 7,000 short Internet texts were extracted from websites such as news network, wechat, blog and Tieba, and the emotion polarity labels of corresponding texts were given. The emotion was divided into three categories: neutral and negative. They correspond to labels (0,1,2). Each news item is represented by a unique ID. Data sets are saved in csv file format The project execution process: data collection -> Input data -> Data preprocessing -> Training and testing model -> Model evaluation Project members: Zhou Maoxi, Dong Yuxi, Chong Yang, Tong Jie Instructor: Wei Xiaomei Setting: Huazhong Agricultural University