In the text literature, many Bayesian generative models were proposed to represent documents and words in order to process text effectively and accurately. As the most popular one of these models, Latent Dirichlet Allocation Model(LDA) did great job in dimensionality reduction for document classification. In this paper, inspiring by Latent Dirichlet Allocation Model, we propose LDCM or Latent Dirichlet Category Model for text classification rather than dimensionality reduction. LDCM estimate parameters of models by variational inference and use variational parameters to estimate maximum a posteriori of terms. As demonstrated - - by our experimental results, we report satisfactory categorization performances aboutour method on various real-world text documents.
展开▼