Automatic Recognition Of Problem Types Of Math Word Problems In Primary Education

With the rapid development of information technology, artificial intelligence technology has become more and more widely used in all areas of life. The research of intelligent solution has also been paid more and more attention by researchers. The research of machine application has made some achievements, and the breadth and depth of research have been strengthened.It is the basic task of mathematical problem-solving theory to find the characteristics and princples of the subject. Primary school mathematics in the application of the problem is troublesome since the wide range of topics, flexible question asking styles. And it's the most critical step to analyze the relationship between entities. So we classified the problems manually with designed labels, then a classification procedure is needed to distinguish them.The classification of machine learning is to reflect the correctness of the problem representation, reflected in the learners, it shows the problem of different types of topics of knowledge and understanding of the use of knowledge. The main research of this paper is the prerequisite and the basic role of the intelligent answer primary school mathematics application, and the efect of the topic classification is directly related to the accuracy of the machine solution.First of all, the thesis systematically expounds the general process and classification model of text classification, and focuses on how to classify the classifier's thought theory and core method.Secondly, the text preprocessing and vector representation of the implied relation problem of primary school mathematics application are explained, including the pretreatment process of the subject matter, the pretreatment process of the implied relation type, the Chinese word segmentation, the filter stop word, the feature extraction, and the vectorization of text, so that the subject text is more easily handled by the classifier. The proper text representation has a great influence on the efficiency and accuracy of the classifier.Thirdly, the method of multi-class selection and the construction of the classifier model are analyzed in detail The initial feature set is obtained by word segmentation and filtering of the title text. The feature extraction method and the statistical method based on word frequency are used to calculate the frequency, The TF-IDF function is used to weight the text feature, and the vector text is represented by the VSM space vector model Then a classifier model is trained then tested providing the VSM space model.Finally, the classification results of the subject text are analyzed statistically, and the F-measurement is used to evaluate the accuracy and efficiency of the classification of the application subject text.