A Research On Automatic Classification Of Mongolian Text

With the development of Mongolian information technology and the release of the international standard of Mongolian coding,emerged a considerable amount of electronic texts in Mongolian.The manual processing of these texts is both time-consuming and back-breaking.This study constructed an automatic text classification system based on supervised learning methods such as Bayesian,Support Vector Machine and Neural Network,and compared the classification performance of these algorithms.This paper falls into three parts:introduction,main body and conclusion.The introduction part introduces the basis of topic selection and research significance,research overview,research data and procedures.Chapter 1 presents the work of text preprocessing such as text cleaning,lemmatization,removal of stop words and feature selection.Chapter 2 elaborates the principle of Bayes algorithm and the Mongolian text automatic classification experiment based on Bayes algorithm.Chapter 3 elaborates the principle of Support Vector Machine algorithm and the Mongolian text automatic classification experiment based on Support Vector Machine algorithm implemented in this study.In Chapter 4 discusses the principle of Neural Network algorithm and the network model structure used in this study and introduces the Mongolian text automatic classification experiment based on this model.The conclusion part summarizes the whole research process and the results of three supervised machine learning methods.Besides,the limitation of the work along with future orientations are pointed out.