| In book recommendation systems,recommendation algorithms play an important role in recommending user preferences and evaluating system performance.In recent years,with the vigorous development of the Internet service industry at home and abroad,such as e-commerce,Douyin small videos and news consulting applications,recommendation algorithms are used to provide services to users.At the same time,recommendation algorithms have attracted more and more attention from researchers,and many methods and indicators of recommendation algorithms have been proposed.At the same time,through the research on collaborative filtering technology and recommender system,we found that the problems existing in the implementation process of collaborative filtering technology mainly include scalability,data scarcity and cold start problems.After investigating a large number of research literature,this thesis proposes an improved personalized recall algorithm LFM to apply in the book recommendation system.The main research work of this thesis is to design a personalized book recommendation system based on the improved personalized recall algorithm LFM.In order to improve the prediction accuracy of LFM model on sparse matrix,an improvement was made on the basis of the original algorithm.The improved personalized recall algorithm LFM was combined with the gradient descent method in deep learning,and the mean shift clustering algorithm was integrated to improve the accuracy and diversity of book recommendation.Finally,the book recommendation system is integrated with the improved personalized recall algorithm LFM,which uses the user’s historical behavior data to calculate the books that the user may like to complete the recommendation work.The system also supports real-time recommendation work by using the historical browsing data generated by the user.At the same time,aiming at the problem of cold start,the idea of timing task is used in system implementation to count the popular book information,and when new users enter the system,the popular book information is recommended to users.In order to evaluate the effect of the improved personalized recall algorithm LFM,root mean square error(RMSE)and mean absolute error(MAE)were adopted as the evaluation indexes of the recommendation results,and book-crossing data set was used for experimental analysis.Comparing the collaborative filtering algorithm,hybrid algorithm,LFM algorithm and the improved LFM algorithm,the comparison experiment proves that the improved personalized recall algorithm LFM proposed in this thesis has a low RMSE in book recommendation,and the recommendation accuracy is obviously better than LFM algorithm and other collaborative filtering algorithms.In the aspect of book recommendation system design,Spring Cloud framework is used to design a book recommendation system with micro-service architecture,and the function and design of the system are introduced in detail in the fifth chapter.Finally,relying on the new Internet technology,the development of a highly available modern book recommendation system,and the sixth chapter introduces the final implementation of the system,and the display effect of the system is shown and analyzed,through the test of the normal operation of each function. |
