Feature Extraction Of Elementary Mathematics Based On Knowledge Graph And Its Application

Knowledge graph technology is applied in different industries,including the educa-tion,and mathematics is the focus of teaching.The subject of mathematics has a strong structure and constitutes a system between mathematical knowledge points.This paper proposes a method for constructing a knowledge map of mathematical knowledge points,and applies the knowledge map to the study of knowledge point feature extraction of math-ematical problems,and finally realizes mathematics Knowledge point feature automatic extraction system,the main work of this paper is as follows:Constructed the basic concept knowledge map and the mathematics knowledge point system map.The basic concept map is constructed according to the mathematical entities involved in the mathematical language and the relations between the entities.There are 551 entities and 561 basic relations.The TransE model is used to train the entity rela-tions to obtain the vector representation of the entity relations,which will be used in the subsequent system.The basis of mathematical semantics understanding and knowledge point relevance scoring.Then,a method for constructing a knowledge graph based on the characteristics of mathematical knowledge points is proposed.Four types of nodes and three types of node relations are defined as the basic structure of the entire knowledge point system and hierarchy.Finally,a mathematics knowledge graph with 683 knowledge point nodes and 741 relations is obtained.Design and realize the automatic extraction system of mathematical knowledge points,which is divided into rule knowledge point extraction and topic knowledge point extrac-tion modules.First,compile the problem-solving rules according to the mathematical problem-solving method,and then mark the rule knowledge points with the knowledge point marking tool.Rule knowledge tagging tool,match the rule knowledge graph with the triples involved in the knowledge element in the knowledge system to obtain multi-ple matching knowledge elements,and then use the trained entity relationship vector to calculate the cosine similarity of the knowledge elements.A total of 507 rules were writ-ten and marked.The topic knowledge point extraction module converts the input topic into a knowledge map through natural language processing,compares the rule knowledge map in the rule library,solves the problem,and then obtains the knowledge point of the mathematical problem according to the rules used in the topic.Finally,use mathematics questions to test the system.The modules of the system can be coordinated to extract the knowledge point characteristics corresponding to the questions.The knowledge point feature extracted by the knowledge point extraction system constructed in this paper is obtained from problem solving,and the knowledge point is defined in the constructed knowledge point feature knowledge graph,so that the extracted knowledge point is more accurate and explanatory.It can also display related knowledge points according to the knowledge map of the knowledge system,so that students can better grasp the knowledge points from the questions and learn.