Research on Testing Text Classification Based on the Sequential Minimal Optimization by Support Vector Machine Method

There are a lot of testing or measuring data would be generated during test or measuring, how to classify the data to further analysis is a very important research appoint. Text classification technology is one of the research priorities in the field of information science, and support vector machine (SVM) has an obvious advantage in solving the problem on text classification. In this paper, according to analyzing the characteristics of English text, we study the whole process of English text classification, build a specific method by improved SVM, and design a program which is suitable for English text classification. Improved sequential minimal optimization (SMO) algorithm is used as the basis of binary classification, and the multi-class classification is realized by using the one-to-one classification method. Simulation results show that our classification model is suitable for different kinds of English texts, and it consistently achieves good performance on text classification tasks. Evaluation of the various indicators on classification can reach bigger than 90%.