Learning Path Recommendation System In Computer Domain Based On Multi-dimensional Knowledge Graph

With the accelerated development of intelligent education,offline education has been unable to meet learners’ knowledge acquisition.Therefore,learners begin to make full use of resources on the Internet for online learning.However,in e-learning,the increasing number of learning resources makes it difficult for learners to find suitable learning resources.This situation leads to a series of problems such as an unsystematic learning process and low learning efficiency.Based on the above problems,the thesis designs a learning path recommendation system to help learners organize learning paths and improve their learning efficiency.Combining the existing learning path recommendation methods,the thesis uses knowledge graph to integrate the computer domain knowledge,and applies the characteristics of learners and learning resources to recommend paths for learners.The main research contents are as follows:1)The computer science knowledge graph(CSKG)is built.The computer science knowledge graph mainly has two levels: course level and knowledge point level.The paper divides the courses into three dimensions: theoretical basis,algorithm/framework,and project practice.According to the expert experience,the prerequisite courses of each course are marked to construct the multi-dimensional course knowledge graph.The knowledge graph can help learners effectively understand the learning overview,and enable them to logically organize and remember knowledge.On the basis of the course knowledge graph,the knowledge graph of knowledge points is constructed so that learners can learn the corresponding knowledge points in the course more carefully.According to the semantic relationship between courses and knowledge points,the two graphs are linked to build the computer science knowledge graph.2)A learning path recommendation method between courses based on multi-dimensional knowledge graph is designed.In order to more accurately capture learners’ preferences,the paper optimizes the neighbor selection rules in the neighborhood aggregation operation of the knowledge graph convolution network(KGCN)algorithm and adds the correlation strength of relationships to entities.The importance of course is calculated according to the attributes of course entities in the knowledge graph.Finally,we propose a learning path recommendation algorithm which combines learner’s preferences and the importance of learning resources to provide learners with the most suitable learning path.3)A learning path recommendation method between knowledge points based on knowledge graph is designed.We calculate the importance of knowledge points by adjusting the corresponding weights according to the characteristics of knowledge points in the knowledge graph.The neural cognitive diagnosis model is used to calculate the learners’ mastery of knowledge points,and the path generation rules are set to generate the learning path of knowledge points.According to the learners’ mastery of knowledge points,we adjust the learning path and recommend the optimal path for learners.4)A learning path recommendation system is built.Learners can query computer domain resources through the system.Based on the knowledge graph visualization technology,learners can also view other resources in the knowledge graph that are associated with the learning objectives.Learners can input learning objectives to obtain corresponding recommended paths.At the same time,learners can test their mastery of knowledge through the knowledge evaluation function.