Personalized Learning Recommendation Model Based On Knowledge Map And Improved Collaborative Filtering

With the rapid development of Internet technology,Online learning resources are beginning to proliferate,it leads to more diversified online learning methods and educational means,but abundant online learning resources are likely to cause the phenomenon of "information overload".When students learn knowledge content through online learning resources,they are prone to "knowledge trek".Therefore,research on the field of personalized learning recommendation is becoming a hot spot.First,This thesis proposes a knowledge map of secondary structure.This structure can guide the learners to fully understand the knowledge system and internal logical relationship of the course.Learners can plan their own learning path according to the learning scheme.The secondary structure includes: a first-level knowledge module and a second-level knowledge point,and the knowledge module contains knowledge points.The structure has the characteristics of simplicity,usability and scalability.Secondly,based on the network structure of knowledge map,the degree centrality of nodes is introduced to calculate the importance of knowledge points in the network,which is one of the recommended basis.Then,this thesis proposes a collaborative filtering algorithm PAD-CF based on learning situation.Compared with the traditional collaborative filtering algorithm,the PAD-CF algorithm improves the accuracy of calculating the similarity of learning situation and learning characteristics between learners,and improve the final recommendation effect.The algorithm optimizes the similarity method,based on the Pearson similarity formula,Three factors related to learning situation and learning characteristics are introduced for improvement,and combined with the K nearest neighbor algorithm and the prediction formula to form a collaborative filtering Algorithm PAD-CF based on the learning situation.The algorithm starts from the cognitive level and learning characteristics of the learners in the course,and calculates the degree of similarity between different learners through PAD-CF to improve the accuracy of the similarity calculation results.Through the correction factor,the difference of difficulty coefficient of knowledge points is reduced,and reduce the impact of the difference on the accuracy of similarity.Finally,the secondary structure knowledge map is fused with the collaborative filtering algorithm PAD-CF based on learning situation,it is used to build personalized learning recommendation model LS-PLRM.Based on this model,it provides learners with a personalized learning scheme that adapts to their learning situation and learning characteristics,and guides learners to learn in depth according to the learning path and the recommendation of knowledge points.Through experimental comparison and verification,learners who learn through the personalized learning recommendation scheme provided by the LS-PLRM model have different levels of improvement compared to learners who use other models for learning,and the learning effect is more obvious.