Research On Construction Of Junior High School Mathematics Knowledge Graph And Its Application In Personalized Recommendation

With the continuous development of big data,artificial intelligence and other advanced technology,the level of information is also constantly improving,and the source of knowledge will no longer limit to traditional textbooks.The proposal of the Internet+education model has given rise to a large number of online learning and education platforms,these platforms provide learning services for users by integrating a large number of high-quality learning resources,which not only provide convenience for learners but also bring troubles to learners such as "information overload" and "knowledge confusion",it also leads to changes in the organization of knowledge and resources.As an important form of information representation,knowledge graph can not only effective for knowledge organization,articulate knowledge nodes and their related relationship,but also help to improve and explain the recommendation results when it is used as auxiliary information in the field of personalized recommendation.Based on the above analysis,this paper builds a junior middle school mathematics knowledge graph-based personalized learning resource recommendation model by combining deep learning and knowledge graph.The main work of the research is as follows:(1)The mathematics knowledge graph of junior middle school has been constructed.To solve the problem that there is no publicly available middle school mathematics knowledge graph in the current education field,we have study the theories related to the knowledge graph,and the middle school mathematics knowledge graph was successfully constructed under the guidance of relevant experts by using a semi-automatic method.The resource of the knowledge graph is the middle school mathematics textbooks and syllabus,and the technology used in the constructing process are OCR,entity labeling,and so on.(2)A personalized learning resource recommendation model has been constructed and coded.Because of the problems existing in the field of personalized learning resource recommendation,the personalized learning resource recommendation is divided into two steps,namely knowledge positioning and resource recommendation.In the part of knowledge positioning,the deep learning algorithm and graph search are combined according to the needs for resources of learners.In the resources recommended module,we have analyzed the learner’s preference for resources,and the feature of resources and the learners are encoded to train the SVM to do the classification,and also the recommended resources fall within the scope of the knowledge identified in the previous step.At the same time,considering the problem that junior middle school students’evaluation of learning content is arbitrary,so in the part of data processing,a method of using implicit feedback to modify the learner’s rating is proposed.(3)The simulation experiment is carried out on the constructed model.The online learning platform is used to simulate the online learning of learners and generate a data set.The collected data set and the knowledge graph constructed in this study are used to train the model.The effectiveness of the personalized learning resource recommendation model is verified by comparing the recommendation results provide by experts and our model.