Research On The Identification And Application Of Concept Prerequisite Relations Based On Subject Knowledge Graph

In recent years,Massive Open Online Course(MOOC)and Coursera have been widely used with the rapid development of Massive Online learning platforms.Based on this,the online learning platform of students’learning process,problem records and other data can be obtained on a large scale.As a carrier of knowledge representation,knowledge graph can effectively organize and describe various knowledge points in course learning.The knowledge graph based on subject knowledge points provides the knowledge base for individualized teaching and efficient individualized learning of students.The construction of the subject knowledge map depends on the extraction of knowledge points and the relationship between knowledge points,including the important relationship between the sequence of courses,which is of great significance to students’ personalized learning.This paper studies the identification of the relationship between the sequence of repair of knowledge points,and based on this,combined with the learning data of students on the online learning platform,carries out personalized learning path recommendation.The work and innovations of this paper are as follows.First of all,this paper studies the automatic identification of the relationship between knowledge points.This paper combines Wikipedia data and MOOC learning resources,and proposes a knowledge point revision relationship recognition that integrates graph embedding.The vector representation of knowledge points is learned from the created Wikipedia directed graph using graph embedding technology,and the semantic and structural correlations between knowledge points are represented based on the cosine similarity between vectors.Through the pre-trained Siamese network,the vector representation of the course knowledge points learned by the improved LDA model is used to obtain the similarity of knowledge points related to the topic distribution.Finally,by fusing some artificial features,and using a binary classifier to identify the relationship between knowledge points in MOOCs.And experiments are carried out on two MOOC datasets to verify the effectiveness of the performance of the proposed method.Secondly,this paper studies the personalized learning path recommendation,and proposes a personalized learning path recommendation algorithm based on the relationship between successive courses and students’ learning status.The algorithm constructs a student relationship graph based on the students’ learning data on the online learning platform,and uses the corresponding graph embedding algorithm to obtain similar student groups from the perspective of graphs.With the help of the knowledge tracking algorithm,students are used to obtain the knowledge point set that they do not fully grasp,and the candidate learning path is searched through the depth-first algorithm with the help of the successive revision relationship between the knowledge points.Finally,the pros and cons of the candidate paths are judged by the designed features.In order to measure the performance of personalized learning path recommendation,a special offline evaluation index is designed.Finally,the superiority of the method proposed in this paper is verified by experiments using the MOOCCubeX dataset.Finally,the application and visualization front-end of the algorithm proposed in this paper in the personalized learning platform based on subject knowledge graph are introduced.