Research And Implementation Of Semi-automatic Construction Technology Of Knowledge Graph In Elementary Mathematics Problems

As the Internet enters a period of rapid development,people's clothing,food,housing,travel and other aspects have undergone rapid changes,and the education industry has gradually changed from traditional mechanized education methods to modern intelligent education methods.Knowledge Graph is the product of the era of big data.It not only can conveniently view the structural relationship of subject knowledge points,but also provide a visual concept for natural language understanding – relational representation and deep semantic reasoning.Therefore,it is of great theoretical significance and practical application value to construct the elementary mathematics Knowledge Graph and guide students to learn independently and give full play to subjective initiative.Based on the analysis of the characteristics of the knowledge of elementary mathematics,this thesis proposes a semi-automatic construction method of elementary mathematics Knowledge Graph based on core concepts and relationships.The main research contents are as follows:Through the analysis and research of natural language processing and knowledge representation methods in the general domain,the natural language processing model and knowledge representation method oriented to mathematics are proposed.According to the different use scenarios of mathematics,the mathematics subject knowledge is divided into four categories: geometry,algebra,probability statistics and text questions.The support vector machine is used to realize the classification of the scenes of the elementary mathematics,which reduces the complexity of the relationship extraction task and facilitates the modular management of knowledge points.Using pattern matching,syntactic analysis and deep learning methods for relation extraction,and fusion dependent syntax analysis and TextRank algorithm,an automatic extraction method of elementary mathematical text relations based on keyword localization is proposed,which realizes the construction of elementary mathematics knowledge graph based on core concepts and relationships.The core concepts and relationships of elementary mathematics constructed by hand are 663.The method of this thesis automatically extracts 856 entities and 22578 triad relations,which enriches and perfects the elementary mathematics Knowledge Graph.Finally,using the constructed mathematical Knowledge Graph to assist the understanding of elementary mathematics,through the analysis of the results of the meaning of the problem,the established mathematical Knowledge Graph can improve the pass rate of the understanding of mathematical meaning,and has a good use value.