Research And Implementation Of Educational Knowledge Graph Management System For Knowledge Structure

In today's society,online teaching has gradually become a new means of teaching.In the learning stage of middle school,students' self-consciousness is poor and independent learning ability is not strong.For junior middle school mathematics,the formation of students' cognitive level of subject knowledge and cognitive structure is also in a critical period.Students often rely on Internet resources to assist their self-study.However,in the era of big data,the teaching resources on the Internet are characterized by a large number of duplicate resources,poor knowledge relevance and inconsistent knowledge representation methods.The mathematics subject knowledge map of junior middle school was constructed,and the Neo4 j graph database was used for knowledge storage,and the front-end Echarts framework was used to graphically display the different relationships between junior middle school mathematics knowledge points and knowledge points.This can not only improve students' learning efficiency,but also enable students to quickly grasp the hierarchical structure of subject knowledge and the relationship between knowledge points.In this paper,according to the characteristics of junior high school mathematics subject knowledge,the hierarchical structure of junior high school mathematics subject knowledge,clear the relationship between junior high school mathematics subject knowledge points.From baidu encyclopedia climb above data,the junior middle school mathematics knowledge as well as foreign data and save as a knowledge map,using the summary of the teaching material knowledge and the relationships between knowledge points,as the basis of reference,the use of named entity recognition technique to extract entities and relationships,junior middle school mathematics knowledge map construction and visualization,and develop knowledge map management system for knowledge management.In the process of knowledge map construction,this paper proposes a Chinese character vector optimization method based on location information and combines the deep neural network model Bi LSTM-CRF to perform entity recognition on the original corpus of junior middle school mathematics subject knowledge obtained by Baidu Baike.Compared with the traditional named entity recognition model Bi LSTM,the Chinese character vector optimization method based on position information can extract the semantic relation between characters and their adjacent position characters,reduce the error rate of word segmentation,and jointly embed characters according to the frequency information of characters in knowledge corpus.Using the optimized character vector as the input of Bi LSTM layer can effectively solve the problem of lack of word connection in Chinese character vector.Since the classification results obtained by Bi LSTM may be invalid,constraint conditions will be added to the part-of-speech prediction using CRF model to avoid semantic conflicts before and after entities in the prediction results and reduce the error rate of entity recognition.Experiments show that this method has good entity recognition effect.