The development of computer and communications technology has resulted in huge amount of data. The automatic text classification technique has become very significant. Naive Bayes algorithm is based on probabilistic model. It is an effective way to deal with automatic text classification. The main task of this paper is to discuss the theoretical basis of Naive Bayes text classifier and describe the process of using Java language to accomplish the classifier. We can divide the classifier into two parts: the feature extraction and the calculation according to the feature. In the feature extraction part, I use the Chinese word segmentation method and the stop words filtering. In the classification part, I calculate the prior probability, the likelihood function value and the maximum a posterior estimation. During the simple test, the author uses the Sogou laboratory’s text classification corpus as the training set and the test set. During the test, the accuracy is between 39% to 56 %. The results show that there is still room for improvement. The paper also includes the discussion of its improvement methods and wider application.

Cleanse the data of "Compendium of Materia Medica" as the full-text data foundation, with herbs as the center, to construct a Chinese medicine knowledge graph including 6 entity types and various relationship types. Use BiLSTM, CRF, and naive Bayes classification to build a question-answering system based on the Chinese medicine knowledge graph.