Automatic Classification of Chinese Long Texts Based on Deep Transfer Learning Algorithm

Traditional Chinese long text automatic classification methods have the problem of missing analysis information, which leads to a low recall rate. This paper designs an automatic Chinese long text classification method based on deep transfer learning algorithm. The literal similarity is obtained, and the morphemes are processed by weighting and quantization. On this basis, the noise data is eliminated, and the dimensionality of the structural features of the Chinese long text is reduced. A deep transfer learning algorithm is used to count the word segmentation information and minimize the loss function value of the category classifier. The relationship between document characteristics and document categories is explored, and an automatic classification mode is set. Experimental results: The recall rates of the Chinese long text automatic classification method in this paper and the other two automatic classification methods are 77.096%, 66.897%, 66.762%, indicating that the performance of the Chinese long text automatic classification method combined with the deep transfer learning algorithm is better.