A text classification model constructed by Latent Dirichlet Allocation and Deep Learning

In this paper, we proposed a mixed model of text classification constructed by latent dirichlet allocation and deep learning. The model present that a text will be represent as a vector computing by latent dirichlet allocation algorithm, and this vector is probabilistic vector of corresponding topic words space. Then we input these topic vectors into a deep learning framework for computing nonlinear relationship of each vector. Finally, we constructed a text classification system. The proposed model achieves a higher accuracy when compared with other current popular algorithms, such as SVM, KNN and TFIDF.