Abstract:
With the development of artificial intelligence, machine learning development, compared with traditional neural network and other non-linear decision and modeling theory, support vector is used in the field of text information processing to solve the classification problem, because of the simple structure and complete theories. This paper puts forward the research on the emotion classification of the depth learning text in view of the intricacies of the massive text emotion data, the inability to grasp the positive and negative information of the text accurately and accurately. Firstly, we introduce the idea of text emotion classification. Then we introduce the text emotion and TF-IDF weight calculation, and improve the TF-IDF weight and mediate the depth of learning Word2Ve word vector + LIBSVM model to train the Internet text data. Finally, the accuracy of the classification reached 92.28%.