| As the Internet Plus Education models continue to deepen,the online learning system has become a new way for many people to acquire knowledge,but in the context of massive learning resources,there are some problems of homogenization and insufficient personalization in online learning system,so researchers begin to study the combination of personalized learning and information technology,but most of the current traditional online learning platforms use mature recommendation technology such as collaborative filtering algorithm,when the user history data is sparse,the performance of the traditional recommendation algorithm will be limited.Therefore,many researchers have introduced representation learning techniques represented by deep learning,which embeds knowledge graph into a continuous vector space to solve the problems of data sparsity and cold-start.But at present,the research on representation learning is mainly in the field of tourism and medicine,while the research on mathematics teaching is less and the depth of theoretical research is not enough,the mathematics in junior high school is mainly expressed in the way of popular language,and the abstraction of mathematics language in senior high school puts forward higher requirements on thinking ability,which leads to the problem of knowledge deviation in the process of mathematics learning for students in this stage.Based on this,this paper conducts a study of representation learning algorithms and high school mathematics knowledge recommendation systems from the perspective of cognitivist learning theory and relying on knowledge graph representation learning technology.The main research content is as follows:Firstly,this paper firstly analyzes the present situation of on-line learning and kn owledge graph representation learning theory at home and abroad,defines the tradition al recommendation algorithm,knowledge graph representation learning and on-line lear ning,and deeply discusses the cognitive learning theory,CTCL guidance framework(Culture,Technology,Content,Learning)and the guidance of zone of proximal developm ent theory to the research process.Secondly,this paper introduces a representation learning approach to solve the traditional recommendation algorithm data sparsity and cold-start problem,a translation model-based representation learning algorithm for shared variable networks is proposed,in which the prediction problem is treated as a sort problem by means of shared variable networks to solve the problem that the representation learning model is inefficient in dealing with complex relationships.After verification,the improved representational learning algorithm outperforms the classical translation model in the link prediction task,providing a support for the recommendation algorithm process framework.Then,this paper designs the process of recommendation algorithm from the angle of cognitive learning theory,combines the improved representation learning algorithm with the traditional collaborative filtering algorithm,and on this basis solves the problem that the traditional collaborative filtering algorithm only considers the behavior information between users and articles and ignores their own content information.Finally,this paper constructs the knowledge graph of senior high school mathematics as the data base,depending on the recommendation technology of collaborative filtering and representation learning,designs and implements the recommendation system of senior high school mathematics according to the design principle of the recommendation system,and validates the effectiveness of the recommendation method.Finally,the author analyzes the problems of recommendation algorithm and system through the result of questionnaire survey. |
