Research And Implementation Of Personalized Recommendation System For Educational Resources

With the growth of overload information resources,the user's demand for personalized service gradually increasing,the recommendation system in e-commerce,social networks and other fields has become an indispensable technology.However,recommendation system in the field of education teaching is developing slowly.As the recommendation technology becomes more mature,we can predict that recommendation system in the field of teaching will be a large number of use.In this paper,the main work can be divided into the following three points:1.Due to the lack of precision of the weighted Slope One algorithm,and the poor adjustment of sparse,it is proposed to integrate the correlation rule correlation and weighted Slope One.The first,using the association rule Apriori to find the associated items for the target item,and then building the Weighted-slope One algorithm model on the lightweight matrix.New fusion algorithm using the python language contrast experiment on the MovieLens dataset: under different k values,new fusion algorithm AW-Slope one of MAE is below Slope one algorithm and Weighted-Slope one algorithm.2.Because the ItemCF algorithm have cold start problem and the recommended accuracy is not high,the fusion improvement algorithm based on content-based and content-based collaborative filtering algorithm is proposed.at the first,using CB algorithm to calculate the similarity between items,and then on the target items of user behavior records matrix to calculate the similarity between items.finally,doing recommendations.The algorithm was evaluated on the MovieLens data set and showed that the average error rate of MAE was significantly lower.3.The education resource title recommendation system is designed and implemented.CB-ItemCF fusion algorithm was applied to topic recommendation system in education resources?We use the questionnaire statistical system recommendation effect.The data showed that 93.3 percent of people were satisfied with the system's analysis of their results,and 84.4 percent were satisfied with the recommended topic.