Design And Implementation Based On Hadoop And Deep Learning Of The Electric-Dealer Platform

With the rapid development of technology," Internet plus " era has come.More and more E-Business Platforms serve people's daily life.How to provide more stable Internet access and better goods recommendations links has become the priority of the E-Business Platforms.The E-Business Platforms of "e-mall",as a comprehensive E-Business system of the energy industry,was explicitly came up with by the State Power Investment Corporation(SPIC)in the 13th five-year plan of the Supplies and Procurement Corporation,Branch.The introduction of Hadoop technology offers support to the stability of platform services,and the research on recommendation algorithm like Collaborative Filtering,KNN and Deep Learning,etc offers support to the humanization of platform services.In this project,conceptions and constructions start in the aspect of the analysis of the recommendation site of the web page,the date design of the recommendation module,the design of the execution process of different recommendation algorithms,the compilation and cleaning of dates for training,the construction and training of neural network for deep learning,the construction of the Hadoop platform and the application of recommendation algorithm on it,the performance and utility test of the Hadoop platform,the test evaluation of the traditional recommendation algorithm and DP recommendation algorithm in different dimensions.For the platform web page,different columns of the home page,detail page and searching page are analyzed for portable recommendations;the neutral network of deep learning of LSTM+GAN is constructed and reached for recommendation based on algorithm;the Hadoop platform applying modules like HDFS,Hive,Kafka,Spark,etc is constructed for the date storage,the transmission of current date flows,the rapid date analysis;for the test of the Hadoop clusters,passing the function and performance test of big date platform by the China Academy of Information and Communications Technology;the test for algorithm is examined and evaluated from the prospective of accuracy,efficiency and date clusters for training.The E-Business Platforms of "e-mall" realize faster reaction and better user experience through the research and construction on the project.The efficiency of the E-Business Platforms improves to a large extent by Recommendation algorithm and big date platform,and the research and application of the project is of great of significance to the social development and academic research.