Design And Implementation Of Book Shopping Recommendation System Based On Hadoop

The Intemet+ initiative has opened a window for the development of various fields and has provided entirely new solutions.The development of informatization has increased productivity,reduced costs,and made people's lives more convenient.However,as a result of the development of informatization,we have entered the era of big data.Nowadays,the amount of data generated by humans over the past few years has exceeded the amount of data in the past century.Faced with the ever-increasing volume of data,how do data producers push their own production data to the target population and how data consumers can quickly obtain the data they are interested are the problem of "information overload" that we face in present world.The recommendation system can effectively solve this problem.However,the traditional recommendation system is easy to encounter bottlenecks and incompetence in the massive amount of data.As a result,the recommendation result is not ideal.Therefore,the research of recommendation system on the big data platform has practical value and important significance.This article studies the mainstream recommendation algorithms in the academic world and the big data framework-Hadoop.The distributed file management system HDFS is used to realize the distributed management of system files,and the distributed computing component MapReduce of Hadoop is used to implement distributed computing.By using the project-based collaborative filtering algorithm in Mahout components,a book market recommendation system based on big data platform Hadoop is designed and implemented.The details are as follows:First,a browser/server model bookstore system is built using Java Web technology.The system uses j Query and Bootstrap on the front end.The system back end uses the decoupling module to develop.The presentation layer framework uses Spring MVC,the business logic layer framework uses Spring,the data persistence layer framework uses Mybatis,and the classes are managed through Spring dependency injection.The database uses the Oracle database.Through Maven project management tools for project management,using Eclipse development tools for project development,Tomcat server for project deployment.Then,a Hadoop cluster is built to collect the behavior data of the store mall through the data collection module Flume,including user browsing,searching,and purchasing behaviors.The massive data framework Hadoop solves the massive data storage and processing problems of the book mall system,and uses the Hive component storage to clean after cleaning through MapReduce.The data and the data migration tool Sqoop migrates the result data from Hive to a relational database.Finally,the Mahout component in the Hadoop framework is used to implement project-based collaborative filtering recommendation so that the book mall system can increase the personalized recommendation service function.Finally,system function tests and performance tests were performed.The experimental results show that the system has a complete function,can provide a good shopping experience and personalized recommendation services for users,and achieves the desired results.