Research And Application Of The Automatic Classification Of Elementary Mathematical Problems Based On SVM

In the field of education,people's eyes gradually from offline teacher counseling,manual marking the examination paper and problem-solving turned to artificial intelligence based on the Internet intelligence education.An important prerequisite for the realization of this new type of mathematical education idea is to convert the text language.The popular talk is about converting the mathematical statement of human understanding into a pre-defined computer storage knowledge for the next step.These processes include problem solving,paper marking automatically and so on.This premise can also be called natural language processing.The classification problem is the main problem in the process of natural language processing.This paper is divided into two parts,the first part is on the primary mathematics problem tokenization as well as Part-of-Speech tagging,the second part is based on SVM on the primary mathematical problem text classification,and then according to different classification methods,the computer can be converted its presentation to solve the problem.In English,each word has a space between them,but the Chinese is different,all the characters are connected together,so the first step is tokenization.But the mathematical expression contains more symbols with specific meaning,so the general segmentation method does not work.Therefore,we need to construct a special word segmenter for mathematical representation.Similarly,the entities in the expression of mathematical language are different from ordinary language,and the entities of ordinary language are usually describe time,place and name.In mathematical expressions,entities that contain important information are belong to mathematical terms,such as triangles,and equations.So it is necessary to define a specific named entity for the primary mathematical direction and then extract it.In this paper,the conditional random field is used to the named entity.In the elementary mathematics,automatic question solving involves many different category.Before we solve a question,the question need to be classified.Then we excute a corresponding solution method according to its category.To classify the question,firstly we preprocess the Named Entity labeled text,the preprocessing includes removing stop words and establishing the bag-of-words model.Secondly,In this paper,we use the measure of Chi square statistic to choose eigenvectors.So that there is a good reduction in the amount of calculation in using the eigenvectors to reduce the dimension,meanwhile the accuracy of classification can be maintained.Finally,according to the method proposed in this paper,we have construct a system to classify the initial mathematical problems and classify them with SVM.The system can accurately mark the named entity,for the latter to solve the problem and other processing to provide knowledge,while the effective classification of the text can be used to solve the problem or make a reasoning pruning.