Research On Key Technologies Of E-commerce Recommendation System In Cloud Computing Environment

It has become a challenging task that how to get valuable information quickly in the massive data under the background of big data era that the data is exploding as the data structure is complex and changeful.In this context,e-commerce recommendation system arises at the historic moment.It can dig out the different behavior preferences of users and provide personalized recommendation service for different users.At present,there are many problems in the recommendation technology,such as data sparsity,scalability,real-time recommendation and cold start.Cloud computing is a style of computing to powerful computing resources and a huge storage capacity in which dynamically scalable and often virtualized resources are provided as a service over the Internet.The dissertation improves the key technologies of e-commerce recommendation system based on cloud computing technology.The main contents of the dissertation are as follows:(1)Cloud computing technology analysis and research.The design idea and implementation methods about recommendation algorithm based on Cloud Computing were given through the research of Hadoop cluster operation principle and its implementation technologies including HDFS distributed file system,MapReduce programming model and H-Base database.(2)Algorithm research.On the basis of cloud computing technology to improve the design idea of the recommendation algorithm,the algorithm is optimized and improved.And the concrete steps is that the MapReduce programming model is applied to decompose and merge the algorithm which is divided into two parts: the serial computing part and the parallel computing part.This corresponds to a number of MapReduce job processes,each of which consists of a decomposition process of Map function and a merge process of Reduce function.(3)Empirical research.In the view of the fact that Hadoop runs best on Linux system,the Hadoop cluster platform is built on virtual machines.Netflix data set is taken as the empirical research object.The running time of the algorithm,the speedup ratio and the average absolute error are selected to text the optimized algorithm.On this basis,the experimental results are summarized.