Design And Implementation Of E-commerce Personalized Recommendation System Based On Mahout

With the continuous transformation of traditional industries to e-commerce,vertical electricity business,O2O(Online To Offline online offline / online to line)era officially announced,e-commerce companies from the rapid expansion to the present increasingly stable.Competition between companies is becoming increasingly fierce.How to retain the user to become the focus of research.Recommended system based on user behavior habits,to tap the potential needs of users,take the initiative to help users find possible to like the goods.Therefore,this thesis studies the design and implementation of personalized recommendation system based on Mahout,the main work is as follows:This thesis introduces the typical recommendation algorithm,including the content recommendation algorithm,the collaborative filtering recommendation algorithm and the recommendation algorithm based on association rules,and introduces the Hadoop platform and MapReduce framework,and analyzes the basic principle and main algorithm of Mahout.Secondly,it analyzes the demand and non-functional requirements of the recommendation system.On this basis,the overall design of the e-commerce personalized recommendation system is given,including the storage layer,the recommendation algorithm and the presentation layer,and divided into three modules,namely,the log module,Recommended engine module and recommended result storage module.And then gives the database design,the system is more important and commonly used data tables were explained and explained.Based on the definition of the data model and the definition of the similarity algorithm,the implementation of the proposed algorithm is described in detail,and the detailed design and implementation of each module are described.Finally,from the point of view of the recommendation system,the various components deployment methods needed by the test system are introduced,and the function and performance of the system are tested in detail.The test results show that the recommendation system achieves the design goal and satisfies the e-commerce platform Recommended requirements.The system uses the Hadoop distributed platform to handle the logging system,making the system more reliable and efficient.The core part of the system recommended the use of the Mahout system,and the use of collaborative filtering algorithm to complete the recommended calculation,to provide users with a better recommendation service.