Based On The Hadoop Cloud Platform Of Smart Recommend Logistics System Design And Implementation

With the high-speed development of the Internet, Hadoop cloud platform is A kind of open source software framework,it is continuing to improve itself has the higher performance and more stable version, in an age of data oriented, obtained more attention. It is an important open source implementation of distributed parallel programming model MapReduce of Google, which is rich in the service interface, and can be deployed in the thousands of nodes in the cluster, computing business to deal with huge amounts of data. For realizing the algorithm of parallel program, based on the MapReduce programming model, developers only need to focus on their own to solve computing tasks, and submitted custom class in MapReduce model to the platform by corresponding interface, the characteristics for the development and research of cloud computing services, big data business processing with great convenience.The main research work of this article is based on Hadoop cloud platform.In the process of thesis research, in the VMware virtualization server set up four work nodes, on the basis of the small cluster of intelligent recommendation algorithm application research work.This paper for the Hadoop platform deployment configuration, with MapReduce programming model on the basis of implementation of distributed parallel computing programming method made a careful study of the research.The paper studies the logistics business platform of the original information such as customer relationship, to a recommendation system based on Hadoop platform framework, with the method of off-line experiment, experiment research was obtained from the business platform in the Oracle database with the original data, and through the simple data ETL function module, data conversion, data is adapted to based on the MapReduce model of the algorithm.In this paper, the intelligent methods recommended by the collaborative filtering algorithm based on the project, the algorithm is the core of rating matrix constructed from users-project project between the co-occurrence matrix, then based on the co-occurrence matrix to quickly calculate the user’s interest.The basic implementation of the algorithm is relatively simple, and in dealing with the data set on a certain scale efficiency is higher.Study to graphs programming model to realize the algorithm, combining with the logistics business platform for logistics industry enterprise users to provide recommendation service, Hadoop platform for data collection of the fragmentation of recommended results localization problems makes the realization of the algorithm, in order to solve the problem, and the size of the existing platform of data growth analysis and comprehensive analysis of system structure, puts forward the recommendation system using Redis to construct the cache data storage algorithm used layer of co-occurrence matrix, at the same time adjust the program flow of original algorithm, to solve the problem of recommended results localization.In this paper, with the two methods on the multiple evaluation indexes was analyzed.After the adjustment program in using Redis cache co-occurrence matrix of the experimental results show that the method in terms of performance and evaluation index had significant improvement, operation time is more appropriate, recommend can achieve good results, at the same time in the process of the growing scale of data set also has good real-time performance and scalability.